Skip to content
View in the app

A better way to browse. Learn more.
hosang I.T.

A full-screen app on your home screen with push notifications, badges and more.

To install this app on iOS and iPadOS
  1. Tap the Share icon in Safari
  2. Scroll the menu and tap Add to Home Screen.
  3. Tap Add in the top-right corner.
To install this app on Android
  1. Tap the 3-dot menu (⋮) in the top-right corner of the browser.
  2. Tap Add to Home screen or Install app.
  3. Confirm by tapping Install.

Tech

Tech Articles from a wide variety of topics and categories
Show all categories

Apple Glasses Reportedly Launching in 'Late 2027' With These Features

Apple is now aiming to release its first smart glasses in "late 2027," according to the latest word from Bloomberg's Mark Gurman.

Meta Ray-Bans
He previously said that Apple planned to begin shipping the glasses by early 2027, but he said the product has faced development delays.

The glasses will feature "oval-shaped cameras, unique colors, and multiple frame styles," according to Gurman. "Over time, Apple believes the glasses could evolve into a health device and eventually incorporate augmented reality technologies capable of improving how people see," he said, but this technology is likely years away.

According to Gurman's sources, Apple's CEO Tim Cook views the glasses as his "top priority" before he hands the reigns to John Ternus on September 1.

The glasses will compete with products in the $200 to $500 range in the U.S., he said.

Like the Meta Ray-Bans, Apple's glasses will have built-in cameras that let users capture photos and videos. There would also be speakers and microphones for music, phone calls, and notifications announced by Siri, he said.

The glasses could offer turn-by-turn walking directions.

As for build quality, he said Apple is designing its own plastic frames, with the company allegedly testing at least four potential designs:A larger rectangular frame, similar to Ray-Ban's Wayfarers
A slimmer rectangular design, similar to the glasses worn by Apple CEO Tim Cook
Larger oval or circular frames
Smaller oval or circular framesApple is exploring a range of color options, including black, ocean blue, and light brown, and the glasses may have vertically-oriented oval camera lenses, he said.

Meta uses frames from the popular glasses brand Ray-Ban.


Unlike the latest generation of Meta Ray-Bans, Gurman does not expect Apple's first smart glasses to have an in-lens augmented reality display. He does not expect Apple's glasses to gain such a feature for at least a few years.Tags: Apple Glasses, Bloomberg, Mark Gurman
This article, "Apple Glasses Reportedly Launching in 'Late 2027' With These Features" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

New Apple TV and HomePod Mini Are 'Nearly Ready' to Launch, New Siri Remote Also Rumored

New models of the Apple TV 4K and HomePod mini are "nearly ready to go," according to the latest word from Bloomberg's Mark Gurman.


Both devices have been ready "for months," but Apple is holding off on launching them until the more personalized version of Siri is available, he said.

"I am told the hardware for the next Apple TV set-top box and HomePod mini has been done for months and that both devices are already in active use among employees at the company's headquarters in Cupertino, California," wrote Gurman.

If you have been closely following Apple TV and HomePod mini rumors, this is a familiar narrative.

The revamped Siri is finally expected to launch as part of iOS 27, iPadOS 27, and macOS 27, which will be unveiled during the WWDC 2026 keynote on Monday, June 8. Following beta testing, the software updates should be widely released in September, so the new Apple TV 4K and HomePod mini models should be available to purchase by then. In other words, the devices are hopefully around 3-4 months away at the latest.

The current Apple TV 4K was unveiled in October 2022, while the HomePod mini was introduced in October 2020, so there has been a long wait for the devices. Nevertheless, Gurman said "don't expect much" in terms of new features for both devices, aside from newer chips that support the more personalized version of Siri.

The current Apple TV 4K has an A15 Bionic chip from the iPhone 13 series, while the HomePod mini uses the S5 chip from the Apple Watch Series 5.

Earlier rumors claimed the next Apple TV would be equipped with the A17 Pro chip, which is the oldest chip that supports Apple Intelligence. The device is also expected to feature Apple's N1 chip for Wi-Fi 7, Bluetooth 6, and Thread.

Gurman expects the next Apple TV to have a similar design as the current model.

There is one new twist, as he was told that the Apple TV's Siri Remote may be "refreshed in some form," but he did not provide any specific details or guarantee that there will be any outward-facing design changes to the accessory.


As for the HomePod mini, it is expected to use an Apple Watch's S9 chip or newer, but it is unclear if or how that chip would fully support the new Siri powered by Apple Intelligence. Other previously-rumored features for the speaker include the N1 chip, improved sound quality, a newer Ultra Wideband chip, and a red color option.

Apple is also expected to update the full-sized HomePod and release an all-new smart home hub this year, with those devices also held up by Siri.Related Roundups: Apple TV, HomePod miniTags: Bloomberg, Mark Gurman, Siri, Siri RemoteBuyer's Guide: Apple TV (Don't Buy), HomePod Mini (Don't Buy)Related Forums: Apple TV and Home Theater, HomePod, HomeKit, CarPlay, Home & Auto Technology
This article, "New Apple TV and HomePod Mini Are 'Nearly Ready' to Launch, New Siri Remote Also Rumored" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Dutch Authorities Dismantle Botnet Linked to 17 Million Infected Devices

Dutch authorities have announced the takedown of a botnet that enslaved millions of infected devices, including computers, tablets, smartphones, and IoT devices, to carry out malicious attacks. The bot network, per the Dutch Politie and the National Cyber Security Center (NCSC), consisted of at least 17 million infected devices. More than 200 servers located in the Netherlands acted as theView the full article
Hacker News

Observability Course for Beginners: Complete Learning Path for Metrics, Logs, Traces, Grafana, Prometheus, and OpenTelemetry

If you are starting your career in DevOps, SRE, cloud engineering, platform engineering, or application support, one skill will keep showing up again and again: observability.
Not just monitoring.
Not just dashboards.
Not just logs.
Real observability.
Modern systems are no longer simple. Applications run across Kubernetes clusters, cloud platforms, microservices, APIs, databases, message queues, containers, serverless components, and third-party services. A single user request may travel through multiple services before returning a response. When something breaks, restarting a server is no longer enough.
Teams need to know:
What happened? Where did it happen? Why did it happen? Who was affected? Which service caused it? Did the latest deployment introduce it? Are we violating our SLO? Should we roll back, scale, or investigate deeper? That is where observability becomes essential.
An observability course for beginners should help you move from “I can see a dashboard” to “I can understand and debug a production system.”
This guide gives you a complete beginner-friendly learning path for observability. We will cover metrics, logs, traces, Grafana, Prometheus, OpenTelemetry, Kubernetes observability, SLOs, hands-on labs, career paths, certification training, and how DevOpsSchool’s Master in Observability Engineering Certification fits perfectly into this learning journey.
What Is Observability?
Observability is the ability to understand the internal state of a system by analyzing the external signals it produces.
In simple words:
Observability helps you understand what your system is doing and why it is behaving that way.
A production system usually gives you three major signals:
Metrics Logs Traces These are often called the three pillars of observability.
But in real engineering teams, observability also includes:
Dashboards Alerts SLOs SLIs Error budgets Incident response Root cause analysis Application performance monitoring Kubernetes monitoring Distributed tracing Telemetry pipelines Runbooks Postmortems Reliability engineering A beginner should understand this from day one: observability is not a tool. It is a practice.
Tools such as Prometheus, Grafana, OpenTelemetry, Loki, Tempo, Jaeger, ELK, Datadog, Dynatrace, and New Relic help you implement observability. But the real skill is knowing how to use signals to troubleshoot systems and improve reliability.
Observability vs Monitoring: The First Concept Beginners Must Learn
Monitoring tells you when something is wrong.
Observability helps you understand why something is wrong.
Monitoring usually works well for known problems:
CPU usage is high Disk space is low Server is down Memory usage crossed a threshold Application returned 500 errors Observability helps with unknown or complex problems:
A payment request is slow only for one region A deployment increased latency for one API endpoint A Kubernetes pod is healthy but users still see failures A database query is slow only under specific traffic patterns A downstream service is creating cascading timeouts Logs show errors but the real issue started in another service Monitoring is still important. You need it.
But monitoring alone is not enough for cloud-native systems.
Observability gives engineers the context needed to debug modern applications.
Why Observability Is Important for Beginners
If you are new to DevOps or SRE, observability may feel like an advanced topic. But it is actually one of the best places to build real production understanding.
Why?
Because observability teaches you how systems behave after deployment.
Many beginners learn Linux, Git, Docker, Kubernetes, Jenkins, Terraform, or cloud platforms. These are excellent skills. But eventually, every engineer faces the same question:
“My application is deployed. Now how do I know if it is working properly?”
Observability answers that question.
It helps beginners understand:
How applications behave in production How infrastructure affects performance How errors appear How latency spreads How Kubernetes workloads fail How alerts are designed How teams investigate incidents How reliability is measured How DevOps and SRE teams make decisions This is why observability is such a valuable career skill.
It connects development, infrastructure, operations, cloud, Kubernetes, monitoring, and reliability into one practical discipline.
Who Should Learn Observability?
Observability is useful for many technical roles.
DevOps Engineers
DevOps engineers need observability to understand what happens after deployment. CI/CD may say a release succeeded, but observability tells you whether production is healthy.
SRE Engineers
SRE engineers use observability to measure reliability, define SLOs, monitor error budgets, respond to incidents, and reduce downtime.
Cloud Engineers
Cloud engineers use observability to monitor cloud infrastructure, managed services, Kubernetes clusters, networking, storage, and application workloads.
Platform Engineers
Platform engineers use observability to build shared monitoring and reliability platforms for development teams.
Developers
Developers use observability to understand how their code performs in production, where errors occur, and which dependencies are slow.
Application Support Engineers
Support engineers use logs, dashboards, traces, and alerts to investigate user issues quickly.
If you work with production systems, observability is not optional anymore.
The Three Pillars of Observability
Let’s understand the foundation.
1. Metrics
Metrics are numerical measurements collected over time.
Examples:
CPU usage Memory usage Request count Error count Request latency Disk usage Network traffic Queue depth Pod restart count Database query duration Metrics are excellent for dashboards, trends, alerts, and SLOs.
Metrics answer questions like:
Is traffic increasing? Is latency getting worse? Are errors rising? Is the service healthy? Which pod is consuming memory? Are we meeting our SLO? Prometheus is one of the most popular tools for collecting and querying metrics.
Grafana is commonly used to visualize those metrics.
2. Logs
Logs are event records generated by applications, servers, containers, databases, and infrastructure systems.
Examples:
Error messages Stack traces Authentication failures API request logs Database errors Deployment events Application warnings Logs are useful when you need details.
Logs answer questions like:
What error happened? What did the application say before it failed? Which user request caused the issue? Which exception occurred? Which dependency returned an error? Popular logging tools include ELK, EFK, Grafana Loki, Fluent Bit, and Fluentd.
3. Traces
Traces show the journey of a request across multiple services.
In microservices, one user request may pass through:
Frontend API gateway Auth service User service Payment service Inventory service Database Cache Message queue Third-party API A trace shows how much time each part took and where the request failed or slowed down.
Traces answer questions like:
Which service caused latency? Which downstream dependency failed? Where did the request spend most of its time? Did the problem start upstream or downstream? Which database query slowed the request? Popular tracing tools include Jaeger, Zipkin, Grafana Tempo, and OpenTelemetry.
Complete Observability Learning Path for Beginners
A beginner should not start by installing every observability tool at once.
That creates confusion.
Instead, follow a layered learning path.
Step 1: Learn Observability Foundations
Start with concepts.
Learn:
Monitoring vs observability Metrics, logs, and traces Telemetry Instrumentation Time-series data Distributed systems Application performance monitoring SLIs, SLOs, and error budgets Incident response Root cause analysis This foundation matters because tools make sense only when you understand the problems they solve.
A beginner mistake is learning Grafana panels without understanding what should be measured. Avoid that.
First learn why observability matters.
Then learn the tools.
Step 2: Learn Metrics
Metrics are the best starting point because they are easier to visualize and alert on.
Learn:
Counter Gauge Histogram Summary Labels Cardinality Aggregation Rate calculation Percentiles Time-series storage Start with basic infrastructure metrics:
CPU Memory Disk Network Then move to application metrics:
Request rate Error rate Duration Active users Queue length Database query time A strong beginner should understand two practical models:
RED Method
Useful for services:
Rate Errors Duration USE Method
Useful for infrastructure:
Utilization Saturation Errors These models help you build useful dashboards instead of random charts.
Step 3: Learn Prometheus
Prometheus is one of the most important tools in modern monitoring and observability.
It collects metrics, stores time-series data, supports powerful querying, and integrates beautifully with Grafana.
Beginners should learn:
Prometheus architecture Scrape model Targets Jobs and instances Exporters Prometheus configuration Prometheus data model Labels PromQL Recording rules Alerting rules Alertmanager Prometheus Operator Kubernetes monitoring PromQL is especially important.
PromQL helps you ask questions like:
What is the request rate? What is the error rate? What is p95 latency? Which pod is using the most memory? Which endpoint is slow? Which service is breaching its SLO? If you want to work in DevOps or SRE, Prometheus is one of the first observability tools you should learn seriously.
Step 4: Learn Grafana
Grafana turns observability data into dashboards, panels, alerts, and operational views.
But Grafana training should not only teach where to click.
A good Grafana learner should know how to design dashboards that help engineers make decisions.
Learn:
Data sources Panels Variables Transformations Dashboards Dashboard folders Dashboard permissions Dashboard provisioning Grafana Alerting Notification policies Annotations Dashboard links Prometheus integration Loki integration Tempo integration A good dashboard answers a real question:
Is the service healthy? Are users affected? Did latency increase? Did the latest deployment cause errors? Which dependency is failing? Are we meeting our SLO? Which logs and traces explain this metric spike? Do not build dashboards for decoration.
Build dashboards for action.
Step 5: Learn Logs
After metrics and dashboards, learn logs.
Logs provide the details that metrics cannot.
Learn:
Structured logging JSON logs Log levels Log aggregation Log parsing Log filtering Correlation IDs Trace IDs Log retention Log cost control Loki or ELK Fluent Bit or Fluentd Good logs are searchable, structured, and connected to traces.
Bad logs are noisy, unstructured, expensive, and difficult to use during incidents.
A beginner should learn how logs support troubleshooting:
A metric shows error rate increased Grafana shows which service is affected Logs show the exact error message Traces show the request path The team identifies the root cause This is how signals work together.
Step 6: Learn Distributed Tracing
Distributed tracing is essential for microservices.
Learn:
Spans Traces Trace IDs Span IDs Parent-child relationships Context propagation Sampling Trace attributes Jaeger Zipkin Grafana Tempo TraceQL basics Flame graphs Service dependency maps Tracing is extremely useful for latency debugging.
For example, if checkout is slow, a trace can show whether the delay happened in payment, inventory, database, cache, or an external API.
This is one of the clearest examples of observability value.
Step 7: Learn OpenTelemetry
OpenTelemetry is becoming the standard way to collect telemetry data.
It helps teams generate, collect, process, and export:
Metrics Logs Traces OpenTelemetry is vendor-neutral, which means your application does not need to be tied to only one observability vendor.
Learn:
OpenTelemetry architecture APIs SDKs Auto-instrumentation Manual instrumentation OpenTelemetry Collector Receivers Processors Exporters OTLP Semantic conventions Context propagation Metrics pipeline Logs pipeline Traces pipeline Kubernetes deployment OpenTelemetry is especially useful when you want to send telemetry to multiple tools such as Prometheus, Grafana, Jaeger, Tempo, Loki, ELK, Datadog, Dynatrace, or New Relic.
For beginners, the best way to learn OpenTelemetry is through hands-on labs.
Instrument one small application.
Send traces to Jaeger.
Send metrics to Prometheus.
Visualize them in Grafana.
Then add logs and correlation.
That is how the concept becomes real.
Step 8: Learn Kubernetes Observability
Most modern DevOps and SRE teams work with Kubernetes.
Kubernetes observability is a must-have skill.
Learn how to monitor:
Nodes Pods Containers Deployments Services Namespaces Ingress Persistent volumes Resource requests Resource limits HPA Cluster events Control plane components Application workloads Important tools include:
Prometheus Operator kube-state-metrics Node exporter Grafana dashboards Loki or ELK OpenTelemetry Collector Jaeger or Tempo Alertmanager Kubernetes observability helps answer:
Why is my pod restarting? Why is the service unavailable? Which namespace uses the most CPU? Are pods under-provisioned? Are resource limits too low? Is autoscaling working? Did a deployment cause the issue? For DevOps and SRE engineers, this is where observability becomes daily work.
Step 9: Learn SLOs, SLIs, and Error Budgets
Observability should not stop at charts.
Mature teams use observability to measure reliability.
This is where SRE concepts matter.
SLI
A service-level indicator is a measurement of service behavior.
Examples:
Availability Request success rate p95 latency p99 latency Data freshness Error rate SLO
A service-level objective is a reliability target.
Examples:
99.9% availability 95% of requests complete under 300 ms Error rate stays below 1% Error Budget
An error budget defines how much unreliability is acceptable.
If your SLO allows 0.1% failure, that 0.1% is your error budget.
SLOs help teams make better decisions.
Instead of asking, “Is CPU high?” you ask, “Are users affected?”
Instead of asking, “Should we deploy?” you ask, “Do we still have enough error budget?”
This is how observability becomes reliability engineering.
Suggested 30-Day Observability Learning Plan
Here is a practical beginner roadmap.
Days 1–5: Observability Foundations
Learn:
Monitoring vs observability Metrics, logs, traces Telemetry Instrumentation Incident response SLO basics Goal: Understand the language of observability.
Days 6–10: Prometheus Basics
Learn:
Prometheus architecture Scraping Exporters Targets Labels PromQL basics Alerting rules Goal: Collect and query metrics.
Days 11–15: Grafana Dashboards and Alerts
Learn:
Grafana data sources Panels Variables Dashboards Alert rules Notification policies Dashboard design Goal: Build useful dashboards and alerts.
Days 16–20: Logs
Learn:
Structured logging Loki or ELK Log parsing Log filtering Correlation IDs Trace IDs Goal: Investigate problems using logs.
Days 21–25: Traces and OpenTelemetry
Learn:
Spans Traces Context propagation OpenTelemetry SDK OpenTelemetry Collector Jaeger or Tempo Goal: Trace requests across services.
Days 26–30: Kubernetes, SLOs, and Capstone
Learn:
Kubernetes monitoring Pod and node metrics SLO dashboards Burn-rate alerts Failure simulation Postmortem writing Goal: Build a complete observability project.
What Should You Learn First: Prometheus, Grafana, OpenTelemetry, or ELK?
This is a common beginner question.
Here is the recommended order:
Observability concepts Metrics Prometheus Grafana Logs Distributed tracing OpenTelemetry Kubernetes observability SLOs and incident response Capstone project Why this order?
Because each skill builds on the previous one.
Prometheus makes more sense when you understand metrics.
Grafana makes more sense when Prometheus has useful data.
Logs make more sense when you can connect them with metrics.
Tracing makes more sense when you understand distributed systems.
OpenTelemetry makes more sense when you already understand metrics, logs, and traces.
Kubernetes observability makes more sense when you understand workloads, services, pods, and telemetry.
This order prevents confusion.
Recommended Observability Training Links with the Right Keywords
Use the following keyword-rich links naturally inside your blog, landing page, or learning content. Each link points to DevOpsSchool’s Master in Observability Engineering Certification because it covers the complete observability learning path: metrics, logs, traces, Prometheus, Grafana, OpenTelemetry, Kubernetes observability, SLOs, assignments, capstones, and certification training.
For Beginners
Start here if you are new to observability and want a complete structured roadmap:
Observability course for beginners
This is the right link for learners who want to understand metrics, logs, traces, Prometheus, Grafana, OpenTelemetry, and Kubernetes observability from the ground up.
For Online Learners
Use this when targeting learners searching for remote or flexible training:
Observability training online
This is useful for professionals who want live, guided, hands-on observability training without depending only on scattered tutorials.
For Certification-Focused Learners
Use this when the user wants a validated learning path:
Observability certification
This is a good fit for learners who want assignments, capstone projects, and certification-based validation.
For DevOps Engineers
Use this when writing for DevOps professionals:
Observability training for DevOps engineers
This is relevant because DevOps engineers need observability to connect deployments, infrastructure, Kubernetes, dashboards, alerts, and production feedback.
For SRE Engineers
Use this when targeting reliability-focused learners:
SRE observability training
This works well for SREs who need SLOs, SLIs, error budgets, incident response, burn-rate alerts, and reliability dashboards.
For Hands-On Learners
Use this when the audience wants labs and projects:
Hands-on observability course
This is a strong anchor because hands-on practice is the fastest way to learn production-style observability.
For Grafana Learners
Use this when discussing dashboards and alerts:
Grafana observability training
This is useful for learners who want to build dashboards, alerts, metrics panels, log views, and trace correlation workflows.
For Prometheus Learners
Use this when discussing metrics and monitoring:
Prometheus training
This is appropriate for learners who want Prometheus metrics, PromQL, exporters, alerting, and Grafana integration.
For OpenTelemetry Learners
Use this when discussing instrumentation and telemetry pipelines:
OpenTelemetry training
This is the right keyword for learners who want to understand OpenTelemetry SDKs, Collector pipelines, traces, metrics, logs, and vendor-neutral observability.
For Kubernetes Learners
Use this when the article focuses on cloud-native systems:
Kubernetes observability course
This is useful for engineers who need to monitor pods, nodes, containers, deployments, services, and Kubernetes workloads.
For Full Career-Focused Training
Use this as the main recommended program link:
Master in Observability Engineering Certification
This is the best anchor text when recommending DevOpsSchool’s complete program for observability engineering, DevOps, SRE, cloud, platform, and application monitoring professionals.
Why DevOpsSchool’s Master in Observability Engineering Certification Is a Strong Fit
A beginner can learn observability in two ways.
The first way is random learning.
You watch one video on Prometheus, one tutorial on Grafana, one blog on OpenTelemetry, one GitHub example for Loki, one Kubernetes dashboard guide, and one article on SLOs. You collect pieces, but you may not understand how everything fits together.
The second way is structured learning.
You start with foundations, then metrics, then Prometheus, then Grafana, then logs, then traces, then OpenTelemetry, then Kubernetes observability, then SLOs, then real projects.
This is where DevOpsSchool’s Master in Observability Engineering Certification fits well.
The program is designed as a complete observability engineering path, not a single-tool tutorial.
It covers:
Observability foundations Metrics, logs, and traces Prometheus PromQL Alertmanager Grafana dashboards Grafana Alerting Loki logs Tempo traces OpenTelemetry OpenTelemetry Collector ELK and EFK Jaeger and Zipkin Datadog Dynatrace New Relic Kubernetes observability SLOs, SLIs, and error budgets Assignments Capstone projects Scenario-based certification exam That breadth matters.
Real companies do not use only one tool.
One team may use Prometheus and Grafana.
Another may use ELK.
Another may use Datadog.
Another may use Dynatrace.
Another may be migrating to OpenTelemetry.
Most cloud-native teams need Kubernetes observability.
A good observability engineer must understand the patterns behind the tools.
The DevOpsSchool certification is a strong fit because it teaches observability as an engineering discipline, not as disconnected software tutorials.
How This Training Helps Beginners
Beginners need structure.
Observability has many tools and terms. Without guidance, it is easy to feel lost.
A structured program helps beginners understand:
What to learn first Why each tool matters How metrics, logs, and traces connect How Prometheus and Grafana work together How OpenTelemetry fits into the stack How Kubernetes changes observability How SLOs connect observability to reliability How to build real projects For beginners, the biggest benefit is confidence.
You do not just learn definitions.
You build working systems.
How This Training Helps DevOps Engineers
DevOps engineers need observability to validate production after deployment.
They need to know:
Did the deployment succeed technically? Did it affect users? Did error rate increase? Did latency increase? Are pods restarting? Are resources under pressure? Are alerts meaningful? Can we roll back with evidence? The DevOpsSchool course fits DevOps engineers because it includes Prometheus, Grafana, Kubernetes observability, OpenTelemetry, logs, traces, alerts, and capstones.
This helps DevOps engineers move from deployment automation to production confidence.
How This Training Helps SRE Engineers
SRE engineers need observability for reliability.
They use observability to manage:
SLIs SLOs Error budgets Burn-rate alerts Incident response Root cause analysis Postmortems Reliability dashboards The DevOpsSchool program fits SRE engineers because it connects observability tools with SRE practices.
SREs do not need dashboards for decoration.
They need dashboards that support reliability decisions.
They need alerts that indicate user impact.
They need traces that identify bottlenecks.
They need logs that confirm root cause.
They need SLOs that guide engineering priorities.
A complete observability course should teach all of that.
How This Training Helps Developers
Developers also benefit from observability.
Modern developers are increasingly responsible for production behavior.
They need to know:
How their code performs Which API endpoints are slow Which database queries are expensive Which exceptions occur in production Which dependencies fail How to add custom metrics How to add trace spans How to write useful structured logs OpenTelemetry is especially valuable for developers because it helps them instrument applications properly.
A developer who understands observability writes applications that are easier to debug, support, and improve.
Practical Capstone Project for Beginners
If you want to prove your observability skills, build this project.
Project: Full Observability Stack for a Microservices Application
Deploy a sample microservices application on Kubernetes.
Then implement:
Prometheus for metrics Grafana for dashboards Loki or ELK for logs Jaeger or Tempo for traces OpenTelemetry for instrumentation Alertmanager or Grafana Alerting for alerts SLO dashboard for reliability Failure simulation Incident report Your dashboard should show:
Request rate Error rate p95 latency p99 latency CPU usage Memory usage Pod restarts Active alerts Error budget burn Related logs Trace links Then simulate failures:
Break one service Add artificial latency Trigger 500 errors Restart pods Increase memory usage Slow down a database query Break an external API dependency Use your observability stack to find the root cause.
This type of project is excellent for interviews because it proves practical ability.
Common Beginner Mistakes in Observability
Mistake 1: Learning Tools Without Concepts
Do not start with dashboards before understanding metrics, logs, traces, and telemetry.
Concepts first.
Tools second.
Mistake 2: Creating Too Many Dashboards
More dashboards do not mean better observability.
A good dashboard should answer a specific question.
Mistake 3: Alerting on Everything
Too many alerts create alert fatigue.
A good alert should be actionable, urgent, owned, and connected to user impact.
Mistake 4: Ignoring Logs and Traces
Metrics show what changed.
Logs show details.
Traces show request flow.
You need all three.
Mistake 5: Ignoring Cardinality
Bad metric labels can create performance and storage problems.
Avoid labels such as user ID, request ID, and session ID in Prometheus metrics.
Mistake 6: Treating Certification as the Finish Line
Certification is useful, but practical skill matters more.
Use certification as a milestone, not the final destination.
Mistake 7: Not Practicing Incidents
You should intentionally break things in a lab.
That is how you learn real troubleshooting.
How to Choose the Best Observability Course for Beginners
Before choosing an observability course, ask these questions:
Does it explain observability vs monitoring? Does it teach metrics, logs, and traces? Does it include Prometheus? Does it include Grafana? Does it include OpenTelemetry? Does it include logs with Loki or ELK? Does it include distributed tracing with Jaeger or Tempo? Does it include Kubernetes observability? Does it teach SLOs and error budgets? Does it include hands-on labs? Does it include assignments? Does it include capstone projects? Does it prepare learners for certification? Does it teach incident response and root cause analysis? If the answer is yes, the course is worth serious consideration.
If the course only teaches one tool, it may still be useful, but it is not a complete observability course.
A complete observability course should help you understand the full production picture.
Final Recommendation
If you are a beginner, observability is one of the best skills you can learn for a DevOps, SRE, cloud, platform, or backend engineering career.
Start with the basics.
Understand monitoring vs observability.
Learn metrics, logs, and traces.
Then move into Prometheus, Grafana, logging, distributed tracing, OpenTelemetry, Kubernetes observability, SLOs, alerts, and incident response.
Most importantly, build projects.
Do not only watch tutorials.
Deploy systems, collect telemetry, create dashboards, trigger alerts, simulate failures, and troubleshoot them.
That is how observability becomes real.
The Master in Observability Engineering Certification by DevOpsSchool is a strong fit for this journey because it brings the complete observability stack into one structured path: Prometheus, Grafana, OpenTelemetry, ELK, Jaeger, Kubernetes observability, SLOs, assignments, capstone projects, and certification validation.
For beginners, it gives direction.
For DevOps engineers, it gives production visibility.
For SRE engineers, it gives reliability skills.
For developers, it gives instrumentation confidence.
And for teams, it creates engineers who can look at metrics, logs, and traces and understand what is really happening in production.
That is the true goal of an observability course.
Not just dashboards.
Not just tools.
Real production understanding.
FAQs
What is the best observability course for beginners?
The best observability course for beginners should cover metrics, logs, traces, Prometheus, Grafana, OpenTelemetry, Kubernetes observability, SLOs, alerts, hands-on labs, and capstone projects.
Is observability hard to learn?
Observability can feel complex at first because it includes many tools and concepts. But if you follow a step-by-step learning path, it becomes manageable.
Should I learn Prometheus or Grafana first?
Learn metrics basics first, then Prometheus, then Grafana. Prometheus collects and queries metrics. Grafana visualizes them.
Should beginners learn OpenTelemetry?
Yes, but after understanding metrics, logs, traces, and instrumentation basics. OpenTelemetry is easier to learn when you understand the signals it collects.
Is observability useful for DevOps engineers?
Yes. DevOps engineers use observability to understand production health after deployments, infrastructure changes, and Kubernetes operations.
Is observability useful for SRE engineers?
Yes. SRE engineers use observability for SLOs, error budgets, incident response, reliability dashboards, alerts, and root cause analysis.
What tools should beginners learn for observability?
Beginners should learn Prometheus, Grafana, OpenTelemetry, Loki or ELK, Jaeger or Tempo, Alertmanager, and Kubernetes observability tools.
What is the role of Grafana in observability?
Grafana is used to visualize metrics, logs, traces, alerts, and SLOs through dashboards and panels.
What is the role of Prometheus in observability?
Prometheus collects, stores, and queries metrics. It is widely used for monitoring, alerting, Kubernetes observability, and SLO measurement.
What is the role of OpenTelemetry in observability?
OpenTelemetry helps generate, collect, process, and export telemetry data such as metrics, logs, and traces in a vendor-neutral way.
Is certification important for observability?
Certification is useful when it includes hands-on practice, assignments, projects, and practical assessment. It helps validate skills and gives structure to learning.
Which certification is good for observability beginners?
A broad certification like DevOpsSchool’s Master in Observability Engineering Certification is a good fit because it covers Prometheus, Grafana, OpenTelemetry, logs, traces, Kubernetes observability, SLOs, assignments, and capstone projects.
View the full article
reporter

Lamine Yamal Teases Upcoming Beats Over-Ear Headphones

Ahead of the upcoming World Cup, football superstar Lamine Yamal has arrived at training camp for the Spanish national team sporting what seems to be the unreleased over-ear headphones that appeared in a U.S. Federal Communications Commission database last week. As suspected, the new headphones are a Beats product rather than an Apple product.


In a post on his Instagram account, Yamal shared several photos and a video clip showing him arriving to training camp with the new headphones in a pink color.



We don't know any other details on the upcoming headphones, and it's unclear whether they are a next-generation version of the Beats Studio Pro or if they will carry a new name. They feature a distinctly different design than the Beats Studio Pro, with flatter exteriors on the ear cups and a completely different headband design that appears to include tubular telescoping arms rather than the wider and flatter arms of the Beats Studio Pro.

A release date for the new Beats headphones is currently unknown, but it shouldn't be too far in the future given that they've already received FCC approval and are being seeded to key influencers like Yamal.Tag: Beats
This article, "Lamine Yamal Teases Upcoming Beats Over-Ear Headphones" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

AirPods Weekend Deals Include AirPods 4 for $99 and AirPods Max 2 for $509

Amazon today has the AirPods 4 available for $99.00, down from $129.00. This is a second-best price on the AirPods 4, which is the base model without Active Noise Cancellation, and it's accompanied by a solid deal on the AirPods Max 2.

Note: MacRumors is an affiliate partner with some of these vendors. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

Amazon provides a June 4 estimated delivery date for free shipping, with faster delivery options for Prime members. We haven't tracked an all-time low price on the AirPods 4 in a few months, so Amazon's deal this weekend is a solid option if you've been waiting for a sale.

$30 OFFAirPods 4 for $99.00

Additionally, you can get the AirPods Max 2 on sale for $509.00 right now on Amazon, down from $549.00. This one is available in Blue and Starlight, with similar June 4 delivery estimates as the AirPods 4. This is a match of the all-time low price on the AirPods Max 2.

$40 OFFAirPods Max 2 for $509.00

Head to our full Deals Roundup to get caught up with all of the latest deals and discounts that we've been tracking over the past week.



Deals Newsletter

Interested in hearing more about the best deals you can find in 2026? Sign up for our Deals Newsletter and we'll keep you updated so you don't miss the biggest deals of the season!




Related Roundup: Apple Deals
This article, "AirPods Weekend Deals Include AirPods 4 for $99 and AirPods Max 2 for $509" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

PAN-OS GlobalProtect Authentication Bypass (CVE-2026-0257) Under Active Exploitation

Palo Alto Networks has warned that a recently disclosed medium-severity security flaw impacting PAN-OS and Prisma Access has come under active exploitation in the wild. The vulnerability, tracked as CVE-2026-0257 (CVSS score: 7.8), refers to a case of authentication bypass that could be exploited by bad actors to set up VPN connections. "Authentication bypass vulnerabilities in theView the full article
Hacker News

Russia-aligned crime group Greyvibe extensively uses AI in attacks

Researchers have uncovered a previously undocumented Russian group that makes extensive use of large language models (LLMs) in its attacks against private, government, and military organizations in Ukraine. It uses a variety of attack vectors along with custom malware, with the goal of intelligence gathering for the ongoing war.
Dubbed Greyvibe by researchers from WithSecure, the group has shown systematic use of generative AI across all stages of its operations, from crafting spear phishing lures and malicious scripts to full on malware development and setting up of backend infrastructure.
“While the activities align with Russian state interests, several observed indicators suggest the group has ties to the broader cybercrime ecosystem, with the group potentially involving current or former cybercriminal actors,” the WithSecure researchers said in their report.
Shifting attack vectors
Greyvibe’s first campaign was launched in August 2025, with a series of spear phishing emails that purported to come from Ukrainian officials and government agencies including the Kyiv City, the Main Directorate of the State Emergency, and the State Service of Special Communications and Information Protection.
The emails included links to ZIP and RAR archives, hosted on Google Drive and a service called 4sync, that contained malware loaders written in Python and JavaScript. The final payload was a custom malware program developed by the group that the WithSecure researchers dubbed PhantomRelay.
In another attack in October, the group experimented with ClickFix-style attacks on fake CloudFlare CAPTCHA pages. These attacks instructed users to open the Windows Run dialog and paste in malicious commands.
Greyvibe also set up fake adult club websites in Ukrainian, as well as fake websites for charities claiming to support the Ukrainian military with FPV drones and UAVs. These attacks distributed several malware programs for both Android devices (FallSpy) and Windows (PhantomRelay and LegionRelay).
The researchers also tracked a website in Russian that they believe was part of the group’s operations; it referenced hard-coded telephone exchange numbers for secure telecommunications that are typically used by the Russian military.
“The intended victimology of this activity remains unclear,” the researchers said. “However, the most plausible hypothesis is that the lure was designed to deceive Ukrainian military personnel by presenting the illusion of access to a Russian military terminal.”
Custom malware developed using LLMs
The PhantomRelay malware program is a remote access trojan (RAT) written in PowerShell that can execute additional custom scripts received from the command-and-control (C2) server. While variants of this program have been observed in activity that might be unrelated to Greyvibe, the group completely rewrote the tool and created a version that was exclusively used in its own operations.
LegionRelay is another PowerShell-based RAT that can similarly execute commands and scripts received from the C2 server; it is used for file enumeration, file exfiltration, screenshot capture, browser data theft, Telegram and WhatsApp data exfiltration, RDP access setup and other actions.
FallSpy is an Android spyware program that can steal contacts, call logs, a list of installed applications, SIM-linked phone numbers, device and network information, Wi-Fi SSID, the phone’s last known location, its public IP address, and media files.
Finally, a series of custom scripts for obfuscating and loading malware was also observed: LOOKVALPS (PowerShell), LOOKVALJS (JavaScript), DAYLIGHT (PowerShell), and TEASOUP (JavaScript).
The WithSecure researchers have determined, with moderate confidence, that several of these custom tools were developed with the help of LLMs. LegionRelay in particular, as well as the background infrastructure serving it, show strong indicators of AI generation. The researchers believe some of the platforms used by the attackers include Ideogram AI, ChatGPT and Google Gemini.
“Greyvibe appears to use AI not only for isolated development tasks, but across multiple operational phases,” the researchers said. “This likely enables the group to compensate for capability gaps, accelerate development cycles, and potentially reduce historical backlinks to prior activity. Given this extensive use, we expect the group’s tradecraft to continue evolving and diversifying, likely increasing the complexity of continuous detection, tracking, and attribution.”
View the full article
CSOonline

Microsoft and security researcher’s dueling posts about cybersecurity disclosures get nasty

Microsoft and a prominent cybersecurity researcher have gotten into a very public and rather personal exchange of unpleasantries about what responsible cybersecurity disclosures should mean in 2026. 
A cybersecurity researcher going by the name Nightmare Eclipse, who has disclosed several cybersecurity holes before patches were available, posted that he had tried to contact Microsoft officials and was rebuffed, which led him to publish details about the bugs.
“When I actively asked you [Microsoft] to communicate with me, you refused, humiliated me and made sure to insult me in front of people. You defame me in public with your CVE-2026-45585 advisory even though you literally deleted the Microsoft account I used to report bugs to you with and I got zero pennies from doing so and I still happily did like an idiot,” the researcher posted, adding that Microsoft has now deleted his GitHub account. “You are proving to everyone that you [are] actively escalating this conflict but I’m done begging you.”
The researcher then made a cryptic threat: “Mark this date July 14th, I will make sure your bones are shattered that day.” 
In another post, the researcher was even more direct: “I was told personally by [Microsoft] that they will ruin my life and they did” adding that Microsoft will “do everything but support the research community, I won’t disclose details, but they sabotage people a lot.”
Microsoft responded with its own post saying that some of the vulnerabilities revealed by the researcher “were not responsibly disclosed” and that there was an “unnecessary risk created by these disclosures,” adding, “uncoordinated disclosures that put proof-of-concept code for unpatched vulnerabilities into the hands of bad actors are never justifiable, and have real-world consequences.”
It was then Microsoft’s turn to get personal, with the veiled implication that the researcher has a bad reputation. “We always have and will continue to welcome vulnerability submissions from  anyone through our public researcher portal, regardless of past interactions or reputation,” the post said.
However, one senior Microsoft security executive posted a slightly more upbeat message, suggesting that the company may now have to rethink how it handles cybersecurity bug reports. 
“At this time, we are not changing our bug bar or the criteria we use to decide when a fix is required, though we will continue to evaluate as conditions evolve. Severity continues to be grounded in real-world impact and exploitability, drawing on the full set of signals in the Security Update Guide,” wrote Tom Gallagher, VP of engineering at the Microsoft Security Response Center (MSRC). 
“We will continue to anchor on a predictable rhythm and a disciplined process, while adapting as needed to the conditions in front of us,” he said. “What we encourage in turn is a thoughtful look at whether the practices that worked well for the patching landscape of a few years ago are still well matched to where the landscape is heading. The fundamentals have not changed. The pace at which they need to be applied is changing.”
CSOonline reached out to both Microsoft and Nightmare Eclipse, and neither provided any clarification or additional comments by publication time.
Frustration on both sides
One of the issues behind the debate over cybersecurity disclosure policies is that many researchers feel that their disclosures are often either ignored or the patch is unreasonably delayed by major vendors, including Microsoft. 
Adding to researchers’ frustration is the fact that vendors often do not communicate well about where things stand with a reported security problem. 
But vendors have their own complaint: they can’t address every one of the many holes that are reported to them quickly, given finite resources, and they must prioritize what they patch.
A related issue is the belief that major vendors, including Microsoft, will quickly prioritize patches once the hole becomes public; one example was the Microsoft Authenticator flaw, which Microsoft had known about for eight years before fixing it after it was publicized.  
Both sides may be right
Consultants and cybersecurity executives said both sides make good points in this instance. 
“Microsoft is right that uncoordinated zero‑day drops create real and immediate risk for customers, and researchers are right that vendors sometimes move only when pushed,” said cybersecurity consultant Brian Levine, executive director of FormerGov. “Both truths can exist at the same time.”
And, Flavio Villanustre, CISO for the LexisNexis Risk Solutions Group, added, “the cry from the security researcher feels like there is something vindictive going on. If the researcher believes that [Microsoft] acted unethically or illegally and has evidence in that respect, they could raise complaints with the appropriate authorities, rather than write a blog post. I am inclined to believe Microsoft more in this case.”
Gary Longsine, CEO of Intrinsic Security, also pushed back against Nightmare Eclipse, questioning whether they are functioning as an objective security researcher.
“This person might have a legitimate grievance of some sort against Microsoft, however, legitimate security researchers don’t do things this way,” he said. “I don’t do things that cause damage to literally billions of innocent bystanders, as retribution for whatever slight I may perceive. This is an attacker, an adversary, not a security researcher.”
Erosion of trust
In addition, Ishraq Khan, CEO of coding productivity tool vendor Kodezi, said that he is concerned about the emotional elements of the exchange between the researcher and Microsoft, because it is eroding trust, and that erosion is potentially the biggest danger.
“The researcher appears to believe the relationship failed long before the disclosures occurred. Reading the public posts, the recurring theme is not simply vulnerability research, but frustration over communication, trust, and access to the disclosure process,” Khan said. “Whether those claims are accurate or not, the researcher clearly believes private channels stopped working and that escalation was the only remaining option.”
And that erosion of trust, Khan said, is a critical issue, because AI, especially autonomous agents, is going to require far more trust between vendors and researchers. 
“The industry is entering a new era of vulnerability discovery. We are seeing increasingly capable AI systems uncover bugs, identify attack paths, and assist researchers in ways that were not possible a few years ago. The volume of discovered vulnerabilities is increasing while the time between discovery and potential exploitation is shrinking,” Khan said. “That changes the dynamics of disclosure. Historically, researchers and vendors were operating on a timeline measured in months. Today, discoveries can spread globally within hours. A breakdown in trust that might have once affected a handful of people can now affect entire ecosystems.”
He added, “the reality is that responsible disclosure only works when both sides believe the system is functioning. Researchers need confidence that findings will be taken seriously. Vendors need confidence that researchers will give them enough time to protect customers. Once either side loses faith in that process, the entire model becomes fragile.”
“What concerns me is that these disputes appear to be becoming more public, more adversarial, and more personal. Once security discussions shift from technical facts to questions of intent, reputation, and motivation, customer protection risks becoming secondary to the conflict itself.”
View the full article
CSOonline

DuckDuckGo's 'No AI' Search Traffic Climbs as Users Reject Google's AI Overhaul

Privacy-focused search engine DuckDuckGo has seen a surge in demand for its "No AI" search option in the wake of Google's May 19th I/O announcements. Google debuted a new "intelligent" search box reimagined with AI. It features AI suggestions as an upgrade to autocomplete, support for follow-up questions, expanded Personal Intelligence for connecting Gmail and Google Photos, and Search agents.


DuckDuckGo told MacRumors that visits to its No AI search page more than tripled after Google's announcement. Traffic hit the 3x mark on May 28th, and has continued to climb. Visits have averaged around 84 percent above baseline consistently since May 19.

DuckDuckGo is embracing demand for No AI search options, and it is promoting new extensions available for Chrome and Firefox that set No AI search as the default.

No AI search has no AI-assisted answers, no chat interface, and it surfaces fewer AI images. DuckDuckGo can be set as the default search engine on Apple devices, but not the specific No AI page. DuckDuckGo has its own AI tools, but they are turned off for people who opt for the No AI experience.

DuckDuckGo plans to add No AI search settings to its original extensions for Chrome, Firefox, Edge, and Opera in the near future.

Along with DuckDuckGo, there are other privacy-focused search engine options that minimize AI results. Paid search engine Kagi is one example, with no visible AI information unless you opt for AI tools. Kagi is $5 per month for a limited number of searches, and $10 a month for unlimited searches.

Because it is a paid search engine, it does not have ads and it does not collect and sell user data.Tags: DuckDuckGo, Google
This article, "DuckDuckGo's 'No AI' Search Traffic Climbs as Users Reject Google's AI Overhaul" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

LG's 39-Inch Ultrawide 5K2K OLED Display Officially Begins Shipping Next Week

Back at CES in January, LG unveiled its UltraGear evo GX9 (39GX950B) display, which it claims is the world's first 39-inch ultrawide 5K2K OLED gaming monitor, offering a large curved canvas in the increasingly popular 21:9 aspect ratio with the added benefit of OLED technology for enhanced contrast with true blacks, standard refresh rates of up to 165Hz, and more.


While LG began taking pre-orders for the UltraGear evo GX9 last month and a few early orders have already trickled out through various channels, LG says that the official kickoff of order shipments starts next week.

LG touts the gaming prowess of the UltraGear evo GX9, but its specs mean it can deliver a premium experience across a variety of use cases, from productivity to media consumption and more.

OLED technology delivers a contrast ratio of 1,850,000:1 across the ultrawide display's 5,120 x 2,160 resolution. At a large 39-inch display size with a 1500R curve, this translates to a density of 143 pixels per inch, which is solid but not enough for true retina-level quality. Still, the large, curved display means many users will often be sitting further from the display than usual to be able to take in the full scope of content on the display, and that should prove plenty sharp in most situations.

The Tandem OLED panel in the UltraGear evo GX9 supports up to 335 nits of typical brightness, which is likely sufficient for most uses but does lag behind some other displays including ones in Apple products. The OLED contrast, color fidelity at up to 98.5% of the DCI-P3 spectrum, and HDR support that can push brightness to 1,500 nits at 1.5% APL and 600 nits at 10% APL should, however, all help to offer a quality viewing experience.


For those who do want to game on this display, the UltraGear evo GX9 features AMD FreeSync Premium Pro and NVIDIA G-SYNC support, as well as 0.03ms response times to keep up with fast-moving content.

On the connectivity side, the UltraGear evo GX9 offers a USB-C port with 90-watt power delivery to a connected computer, as well as DisplayPort 2.1 and HDMI 2.1.

We'll be looking to go hands-on with the LG UltraGear evo GX9 as soon as we can, and we'll report back on how well it works for Mac users, but for now LG is taking orders on its own site priced at $1,799.99, and it's also available at Amazon for the same price with delivery quotes starting around June 8.

MacRumors is an affiliate partner with LG and Amazon. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.Tags: LG, OLED
This article, "LG's 39-Inch Ultrawide 5K2K OLED Display Officially Begins Shipping Next Week" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

iPhone Driver's License Feature Set to Expand to a 15th U.S. State

Apple's digital driver's license feature in the Wallet app is set to expand to Virginia, according to a person familiar with the matter.


In select U.S. states, residents can add their driver's license or state ID to the Wallet app on the iPhone and Apple Watch, and then use it to display proof of identity or age at select airports and businesses, and in select apps. The feature has rolled out to 14 states so far, including Arkansas earlier this week, and it is also available in Puerto Rico.

The other states are Arizona, Maryland, Colorado, Georgia, Ohio, Hawaii, California, Iowa, New Mexico, Montana, North Dakota, West Virginia, and Illinois.

Now, Apple is preparing for the feature to go live in Virginia, the person said. However, we do not have an exact timeframe for availability. Towards the end of 2025, Virginia's Department of Motor Vehicles said it planned to support the Apple Wallet's digital ID feature in the coming months, so hopefully it goes live soon by this point.

When the feature goes live, Virginia residents will be able to set it up by opening the Wallet app on the iPhone and tapping on the plus sign in the top-right corner. Next, they will tap on Driver's License and ID Cards, select Virginia from the list once it is added, and follow the on-screen steps to complete the process.

Apple Wallet IDs are accepted at TSA checkpoints at hundreds of U.S. airports for domestic travel. Given that Apple Wallet IDs are not accepted by law enforcement, and lack many other use cases, carrying a physical ID is still necessary.

If you live in a state that does not yet offer Apple Wallet IDs, you can create a Digital ID based on your U.S. passport, and present it at the same participating TSA checkpoints, for age and identity verification purposes during domestic travel. It is not a replacement for a physical passport, and it cannot be used for international travel.

The passport feature requires iOS 26.1 or watchOS 26.1 and later.Tag: Apple Wallet
This article, "iPhone Driver's License Feature Set to Expand to a 15th U.S. State" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Everything We Know About OpenAI's Planned iPhone Rival

OpenAI is developing a smartphone intended to compete directly with the iPhone, in what appears to be a significant departure from the company's previously stated hardware strategy. Here's everything we know so far.


Analyst Ming-Chi Kuo published his findings in late April following supply chain checks, describing the device as an "AI agent phone" built around a continuous, context-aware interface rather than individual apps. Kuo argued that the smartphone is the only device that captures a user's full real-time state, including location, activity, communication, and context, making it uniquely suited to AI agent inference.

He said fully controlling both the operating system and the hardware is the only way for OpenAI to deliver a comprehensive AI agent service, and that AI agents will fundamentally shift how people interact with a phone, moving the focus from launching individual apps to completing tasks through a seamless interface.

Specifications

OpenAI's phone is said to use a customized version of MediaTek's Dimensity 9600 processor, built on TSMC's N2P node in the second half of 2026. Kuo initially named both MediaTek and Qualcomm as chip partners but has since said MediaTek appears "better positioned to become the sole processor supplier."

Luxshare Precision Industry is believed to be the exclusive manufacturing partner. Separately, Kuo reported that Sunny Optical has secured component orders for two OpenAI devices, including the smartphone. This is likely for the camera module.

The device's headline known hardware specification today is its image signal processor, which includes an enhanced HDR pipeline intended to improve real-world sensing through the camera. It is also said to use two AI processors for handling different tasks simultaneously, such as vision and language processing, along with fast memory and storage and security features to isolate processes.

What About Jony Ive's Devices?

The phone represents a notable reversal in OpenAI's publicly stated strategy. The company's hardware ambitions had previously been described as centered on non-phone form factors developed with former Apple design chief Jony Ive, whose startup io Products OpenAI acquired for $6.5 billion in May 2025. Ive and CEO Sam Altman had specifically said they did not want to build a device with a screen, with Altman describing a prototype to employees as "the coolest piece of technology that the world will have ever seen."

The first product from that collaboration was delayed out of 2026 and has since been identified as a smart speaker with an integrated camera, priced between $200 and $300 and expected to launch in early 2027. Other devices reportedly in development include smart glasses, a smart lamp, and potentially earbuds, though those products are further out on the roadmap and some could be cancelled.

OpenAI has also been aggressively recruiting from Apple's hardware ranks, hiring over 40 former Apple employees. The hires include former Apple designers Evans Hankey, Tang Tan, and Scott Cannon, prompting Apple to offer its iPhone Product Design team retention bonuses of up to $400,000 in restricted stock units to counter the poaching.

Timeline

Mass production of OpenAI's smartphone was originally believed to be targeted for 2028, but Kuo has since revised that expectation to the first half of 2027. The accelerated timeline is said to reflect OpenAI's planned IPO, where a compelling hardware product could strengthen the company's investor narrative, as well as intensifying competition in the AI agent phone category. Kuo projects combined 2027 and 2028 shipments could reach around 30 million units if development stays on track.

What Does It Mean for Apple?

If the broader hardware lineup ships, OpenAI will be a direct competitor to Apple across several product categories. Apple is rumored to be developing smart glasses, AirPods with cameras, an AI pendant, and a smart home hub with enhanced Siri capabilities. On the day Kuo published his initial report, Altman posted on X that it "feels like a good time to seriously rethink how operating systems and user interfaces are designed."Tag: OpenAI
This article, "Everything We Know About OpenAI's Planned iPhone Rival" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

DNS-AID will make AI agents easier to discover, says Linux Foundation

As AI agents become more numerous and more communicative, keeping track of where to find them is becoming increasingly important. Numerous proprietary agent registries are on the market, but the Linux Foundation suggests we simply extend the distributed, open Domain Name System (DNS) infrastructure we already have.
The foundation is now inviting contributions to the DNS-AID project, a standard way for AI agents to discover, verify, and communicate with one another over DNS that requires no new infrastructure. It enables agents and Model Context Protocol (MCP) servers to use DNS as a global, vendor-neutral directory.
While many details remain to be worked out, the proposal suggests domain owners create a new well-known address that can provide a starting point for agents looking for one another: _index._agents.{domain}.
This approach ensures that agent discovery remains scalable, secure, and compatible with the protocols that underly the internet, the Linux Foundation said.
“AI agents are quickly becoming the connective tissue of the modern internet, but without secure, open discovery infrastructure, that connectivity becomes a liability,” said Jim Zemlin, CEO at the Linux Foundation. “DNS-AID helps anchor agent discovery in the DNS infrastructure that the internet already trusts.”
DNS-AID was initially developed by staff at Infoblox, and the latest internet draft of the DNS-AID proposal includes contributions from staff at Deutsche Telekom and Amazon. The Linux Foundation said it intends that DNS-AID will remain vendor-neutral.
This article first appeared on InfoWorld.
View the full article
CSOonline

Windows PC Industry Reacts to Apple's Most Affordable MacBook Ever

A few months ago, Apple released the MacBook Neo, its most affordable MacBook ever. At the time, an ASUS executive admitted that the laptop came as a "shock" to the Windows PC industry, which is now in the process of responding.


Acer today introduced a Swift Air 14 laptop, with U.S. pricing starting at $699. By comparison, the MacBook Neo starts at $599 with a 256GB SSD and 8GB of RAM, or at $499 for college students and educational staff. However, the MacBook Neo costs an equal $699 when configured with a doubled 512GB of storage and a Touch ID button.

Powered by a new Intel Core Series 3 processor, the Swift Air 14 features a 14-inch display with a 120Hz refresh rate and a resolution of 1,920 × 1,200 pixels, up to a 512GB SSD, up to 16GB of RAM, an all-aluminum enclosure, and quad speakers with DTS:X Ultra audio. Like the MacBook Neo, the laptop supports Wi-Fi 6E.

Acer's Swift Air 14
The laptop is equipped with two Thunderbolt 4 ports, a USB-A port, and a 3.5mm headphone jack, and Acer says a 70 Wh battery provides up to 19 hours of battery life for video playback and up to 16 hours of battery life for web browsing.

Like the MacBook Neo, the Swift Air 14 is available in colorful finishes, including sage green, frost blue, blossom pink, and lilac purple.

Acer said the Swift Air 14 will be available in North America starting in August.

Meanwhile, Qualcomm this week announced the Snapdragon C, a new processor designed for "entry-tier laptops" priced at "$300 and up." Qualcomm said the processor delivers "responsive everyday performance" with "breakthrough power efficiency." The first laptops powered by the Snapdragon C are expected to launch later this year, with committed brands including Acer, HP, and Lenovo, according to Qualcomm.

Qualcomm's Snapdragon C processor
Indeed, Acer has previewed the Aspire Go 15, the first laptop powered by the Snapdragon C processor. The laptop will have an "affordable" price point, but Acer did not provide specific pricing or a release date. Key specs include a 15.6-inch display with a resolution of 1,920 × 1,080 pixels, up to a 512GB SSD, up to 8GB of RAM, a 1080p webcam, two speakers, two USB-C ports, one USB-A port, one HDMI port, and a 3.5mm headphone jack.

Acer said the Aspire Go 15 is made from 100% recyclable materials and has some components made from recycled plastic, so it sounds like the laptop will not have an all-aluminum enclosure like the MacBook Neo and the Swift Air 14.

Finally, ASUS commented on the MacBook Neo again during its annual shareholders meeting today. According to Taiwan's Economic Daily News, ASUS's chairman Jonney Shih said that the company can learn from Apple's cost-efficient strategy with the MacBook Neo and views it as an opportunity. Stay tuned, he said.

On an earnings call last month, Apple's CEO Tim Cook said that customer response to the MacBook Neo had been "off the charts" since its launch.

Apple was very optimistic about the MacBook Neo before announcing it, but the company still "undercalled" the level of enthusiasm that the laptop would generate, according to Cook. He said that MacBook Neo demand exceeded Apple's expectations and helped to drive a record number of first-time Mac buyers last quarter.

"We could not be happier with how things are going at the moment," said Cook.

As for the Windows PC industry, perhaps not so much.Related Roundup: MacBook NeoTags: Acer, Asus, Intel, Qualcomm, WindowsBuyer's Guide: MacBook Neo (Buy Now)Related Forum: MacBook Neo
This article, "Windows PC Industry Reacts to Apple's Most Affordable MacBook Ever" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Certifiably random: Swiss researchers claim perfect random number source

Researchers in Switzerland claim to have built a perfect random number generator from two quantum superconducting chips, a 30-meter-long pipe, and some software. The resulting device could be used to generate cryptographic keys, or to offer a “public randomness service” for lotteries or blockchain applications, they say.
They’re not the first to make the claim.
Many sources of randomness are biased. For example, coins or dice tend to favor one side. “Even modern random number generators, which are based on quantum mechanical effects like the reflection of photons from beam splitters, are not entirely immune to such a systematic error or ‘bias’,” said Andreas Wallraff, one of the leaders of the research team at ETH Zurich.
Similar biases can be found in purely software-based pseudo-random number generators. This has led to security problems in IoT devices and WhatsApp, among other applications.
To get around that, the researchers set up of two supercomputing chips, each representing one qubit, cooled to near absolute zero. The chips are connected by a 30-meter-long microwave guide, similarly cooled, and the microwave photons flying between them create a situation of quantum entanglement.
The results produced by this process are then transformed via a special algorithm to generate perfect randomness. “The resulting sequence of zeros and ones is now really perfectly random, and we can even certify that,” said Renato Renner, the other team leader. “The technical improvements allowed us to create random numbers that will remain perfectly random for all eternity.”
The team published their results this week in an article entitled “Experimental randomness amplification” in Nature.
View the full article
CSOonline

Attackers Use LLM Agent for Post-Exploitation After Marimo CVE-2026-39987 Exploit

An unknown threat actor has been observed using a large language model (LLM) agent to conduct post-compromise actions after obtaining initial access following the exploitation of a publicly-accessible Marimo network using a recently disclosed vulnerability. "The attacker compromised an internet-reachable Marimo notebook via CVE-2026-39987, extracted two cloud credentials from the compromisedView the full article
Hacker News

Best Apple Deals of the Week: M5 MacBook Air Gets $199 Discounts, Plus Save on Samsung's New Monitors

The M5 MacBook Air hit new all-time low prices this week, with $199 off nearly every model of the computer on Amazon. We're also tracking an ongoing low price on the AirPods Max 2, plus great discounts from Anker and Samsung.

Note: MacRumors is an affiliate partner with some of these vendors. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

M5 MacBook Air


What's the deal? Take $199 off M5 MacBook Air
Where can I get it? Amazon
Where can I find the original deal? Right here
$199 OFF13-inch M5 MacBook Air (512GB) for $899.99
$199 OFF13-inch M5 MacBook Air (16GB/1TB) for $1,099.99

Amazon has sweetened its deal on the 512GB 13-inch M5 MacBook Air this week, dropping the price of the notebook down to $899.99, from $1,099.00. This is a new record low price on the 13-inch M5 MacBook Air, and you'll find $199 off every 13-inch model right now on Amazon.

Anker


What's the deal? Save on Anker charging accessories
Where can I get it? Amazon
Where can I find the original deal? Right here
$49 OFFAnker Prime 3-in-1 Wireless Charging Station for $109.99

Anker's new Prime 3-in-1 Wireless Charging Station has been marked down to $109.99 on Amazon, down from $149.99. This is one of Anker's newest accessories, and Amazon's sale today is just $5 higher compared to the all-time low price.

Samsung


What's the deal? Save on Samsung's new 2026 monitors
Where can I get it? Samsung
Where can I find the original deal? Right here
$50 OFF PLUS EXTRASSamsung 2026 Monitors

Samsung's newest monitors are now available to purchase this week, including the Odyssey G8, ViewFinity S8, and Movingstyle Essential. All of these are available with a $50 launch discount, plus your choice of extras including up to $300 in Samsung credit on a future purchase, a free Music Studio speaker, or free Galaxy Buds4 Pro.

AirPods Max 2


What's the deal? Take $40 off AirPods Max 2
Where can I get it? Amazon
Where can I find the original deal? Right here
$40 OFFAirPods Max 2 for $509.00

Amazon this week has a record low price on the AirPods Max 2, now available for $509.00, down from $549.00. This sale is available in two colors of the headphones.

If you're on the hunt for more discounts, be sure to visit our Apple Deals roundup where we recap the best Apple-related bargains of the past week.



Deals Newsletter

Interested in hearing more about the best deals you can find in 2026? Sign up for our Deals Newsletter and we'll keep you updated so you don't miss the biggest deals of the season!




Related Roundup: Apple Deals
This article, "Best Apple Deals of the Week: M5 MacBook Air Gets $199 Discounts, Plus Save on Samsung's New Monitors" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Top Stories: iOS 27 Leaks, MacBook Ultra Rumors, and More

WWDC 2026 is coming up very quickly now and we're continuing to learn more about what to expect in iOS 27 and other updates, so make sure to check in to see the latest as we head toward Apple's big week, which kicks off with the traditional keynote on Monday, June 8.


While we may not see anything on the hardware side at WWDC, Apple does have plenty of products in the pipeline, and this week gave us an opportunity to check in on where things stand with the high-end "MacBook Ultra," the long-rumored foldable iPhone, and more, so read on below for all the details!

Top Stories

iOS 27's Siri App and 'Search or Ask' Feature Shown in Leaked Images

With less than two weeks to go until the official unveiling of iOS 27 at WWDC 2026, Bloomberg's Mark Gurman has shared some new re-created screenshots showing off what the revamped Siri will look like, in both standalone app form and a pop-up "Search or Ask" version associated with the Dynamic Island. Additional re-created screenshots show how Siri and AI will be more integrated into the Camera and Photos apps.


The revamped Siri will use a dark color scheme similar to that seen on WWDC 2026 promotional artwork, and iOS 27 will include other enhancements such as revamped AirPods settings, quality improvements for Genmoji and Image Playground creations, and more.

MacBook Ultra: 5 Features That Could Justify the Name

Reports and rumors suggest the next MacBook Pro that Apple will release might not be a ‌MacBook Pro‌ at all. It could actually be something altogether new and more exciting – a "MacBook Ultra" – positioned above the Pro as Apple's top-tier laptop, suggesting that the current M5 Pro and M5 Max models will remain on sale when it launches.


In a recent recap, we listed the key features we are expecting in the MacBook Ultra, which is likely to go on sale either later this year or in early 2027. As things stand, the latter time frame is now looking more likely, owing to the global memory chip shortage.

Apple Watch for Diabetes: The Latest on Apple's Plans for Non-Invasive Blood Sugar Monitoring

For many years now, it has been rumored that the Apple Watch will eventually gain non-invasive blood sugar monitoring capabilities, which would enable millions of people with diabetes to track their blood glucose levels without needing to prick their skin with a needle or wear a dedicated continuous glucose monitor.


According to Bloomberg's Mark Gurman, Apple recently shifted oversight of the project from its platform architecture chief Tim Millet to Zongjian Chen, a senior engineer overseeing advanced technologies within the company. He framed this change as positive news for the project, which has apparently been in development for more than 15 years.

Apple Seeds First iOS 26.6 and iPadOS 26.6 Betas to Developers

Even though WWDC is right around the corner, Apple still has another iOS 26 update in the works to tide us over until iOS 27 is ready for prime time, and that's iOS 26.6, which saw its first beta release this week.


We haven't spotted much new in this update yet other than potentially a new alert that will pop up when you've reached the maximum number of blocked contacts, but with that limit into the thousands, most users won't ever hit the cap.

Ferrari Reveals $640,000 EV Co-Designed by Jony Ive

Despite billions of dollars in investment, the Apple Car never came to fruition, but the just-unveiled Ferrari Luce may offer a glimpse of some things we might have seen had Apple's project panned out.


The $640,000 Luce is Ferrari's first all-electric car, and former Apple design chief Jony Ive and his LoveFrom collective were heavily involved in the design of the vehicle.

First Cases for Apple's Foldable iPhone Surface Online

Foldable smartphones present special challenges for case manufacturers looking to offer protection for the devices while still allowing them to fold and unfold properly, so third-party companies are already hard at work designing options for Apple's upcoming foldable iPhone.


Case makers routinely begin mass producing accessories ahead of a new iPhone announcement, working from dummy units or leaked CAD files to size their molds. Their designs are speculative, but they have historically proven accurate to the millimeter, since accessory makers cannot afford to be left without product on launch day.

Meanwhile, we continue to hear about hiccups as Apple seeks to ramp up toward mass production on the new device, with the latest being that Apple's supply chain is seeing issues with early-stage assembly procedures affecting production yields. This comes after word that issues with hinge reliability were also cropping up. Apple is, however, reportedly still aiming for a release later this year, though supplies may be very limited to start.

MacRumors Newsletter

Each week, we publish an email newsletter like this highlighting the top Apple stories, making it a great way to get a bite-sized recap of the week hitting all of the major topics we've covered and tying together related stories for a big-picture view.

So if you want to have top stories like the above recap delivered to your email inbox each week, subscribe to our newsletter!Tag: Top Stories
This article, "Top Stories: iOS 27 Leaks, MacBook Ultra Rumors, and More" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

First Look at iPhone 18 Pro Color Options Revealed by Dummy Models

Leaker Sonny Dickson today shared images of iPhone 18 Pro dummy models in the device's four rumored colors, offering the first real-world look at what to expect from the lineup visually.


Corroborating previous rumors, the dummies show the ‌iPhone 18 Pro‌ Max in Light Blue, Black, Silver, and Dark Cherry. Dickson said "Cherry will probably be the next hit, orange did very well." Cosmic Orange was the signature color of the iPhone 17 Pro and proved popular with customers.

Dark Cherry is expected to serve as the headline new color for the ‌iPhone 18 Pro‌ models this year. The color has been in the rumor mill since at least February 2026, when Bloomberg's Mark Gurman reported that Apple was testing a deep red finish for the ‌iPhone 18 Pro‌ and ‌iPhone 18 Pro‌ Max. At the time, Gurman described the shade as a deep red, and separate reporting from a Chinese leaker later suggested the color was very likely to make the cut, partly because Android rivals were already prototyping the same shade.



The picture sharpened in April, when Macworld reported that the color would be called Dark Cherry and would be closer to wine than a brighter red, and considerably more muted than Cosmic Orange. The leaker known as "Instant Digital" subsequently corroborated that name, characterizing the shade as a combination of burgundy, coffee, and deep purple. "Instant Digital" has a good track record on Apple color leaks, having accurately predicted the yellow finish for the iPhone 14 and iPhone 14 Plus.

Macworld's reporting also identified the full four-color lineup, with internal Pantone codes said to be in use at Apple: Light Blue (Pantone 2121), described as resembling the current iPhone 17's Mist Blue; Dark Cherry (Pantone 6076); Dark Gray (Pantone 426C); and Silver (Pantone 427C), said to be similar to the current generation.

The latest images are significant because they mark the first time the rumored colors have been depicted in physical, real-world form rather than renders or supply chain descriptions. That said, dummy models are typically made from plastic or low-quality metals and are not finished to the same standard as production units, meaning the tone and saturation of each color could vary from what Apple ultimately ships. With that caveat, the dummies are consistent with the earlier rumors, suggesting that this will indeed likely be the final color palette of the device.

The ‌iPhone 18 Pro‌ and ‌iPhone 18 Pro‌ Max are expected to be announced in the fall alongside the first foldable iPhone.Related Roundup: iPhone 18 ProTag: Sonny Dickson
This article, "First Look at iPhone 18 Pro Color Options Revealed by Dummy Models" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

iPhone 18 Pro's Camera Upgrade Will Cost Apple 50% More

The iPhone 18 Pro and ‌iPhone 18 Pro‌ Max's all-new variable aperture lens will cost Apple 50% more than the camera unit used in current models, according to supply chain analyst Ming-Chi Kuo.


Variable aperture has been one of the most persistent iPhone camera rumors of the past few years. Kuo first flagged the feature in late 2024, and it has since been corroborated by multiple reports and apparently entered production earlier this year.

Unlike the fixed f/1.78 aperture found on every iPhone Pro from the 14 Pro through to the 17 Pro, a variable aperture will physically adjust the size of the lens opening to control how much light reaches the sensor, offering better exposure control and greater flexibility over depth of field.

Kuo said that the component has an average selling price roughly 50% higher than the seven-element plastic lens Apple currently uses in the iPhone 17 Pro's main camera. Sunny Optical set to supply Apple between 40 and 50% of orders

Sunny Optical has also become a new compact camera module (CCM) supplier for Apple, initially producing the camera for the MacBook Neo. ‌MacBook Neo‌ shipments have come in significantly better than expected, with Kuo doubling his 2026 forecast from 5 million to 10 million units, a notable upward revision as the entry-level Mac has materially exceeded early expectations.

Looking further ahead, the 2028 iPhone's ultra wide camera module is expected to move away from flip-chip packaging in favor of an improved COB (chip-on-board) design, with Sunny Optical well positioned to become a supplier at that point. A COB ultra-wide module could be thinner or smaller, leaving more room for other components, or simply deliver better image quality from the same physical footprint.

Beyond Apple, Kuo says Sunny Optical has secured component orders for two OpenAI devices, including a smartphone and a pocket or mobile device.

The ‌iPhone 18 Pro‌ and ‌iPhone 18 Pro‌ Max are expected to launch in the fall alongside the first foldable iPhone. Related Roundup: iPhone 18 ProTags: 20th-Anniversary iPhone, Ming-Chi Kuo
This article, "iPhone 18 Pro's Camera Upgrade Will Cost Apple 50% More" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Notepad++ vulnerabilities could enable arbitrary code execution on Windows systems

Two arbitrary code execution vulnerabilities in Notepad++ let local attackers run commands of their choice on Windows machines by tampering with the editor’s XML configuration files, with both flaws rated High at CVSS 7.8.
The flaws, tracked as CVE-2026-48778 and CVE-2026-48800, affect every version of the editor up to and including 8.9.6, Notepad++ said in a release note. However, the vulnerabilities were patched the same day in version 8.9.6.1, alongside a third lower-severity crash bug, CVE-2026-48770, Notepad ++ author Dun Ho wrote in the release note.
The two code execution flaws share a single design weakness. Notepad++ stores user choices, such as the path to the command-line interpreter and the list of user-defined commands, inside XML files in the user’s profile directory. The editor reads those values and passes them to the operating system as commands without checking what they contain, according to a GitHub Security Advisory on Notepad++ published on May 27.
Anyone who can write to the XML files can decide what the editor executes, the advisory said.
A backdoor that hides in the Run menu
The more concerning of the two flaws, CVE-2026-48800, targets the file that holds user-defined Run menu entries.
Notepad++ reads its user-defined commands from a file called shortcuts.xml and accepts whatever it finds there without validation, the advisory said. An attacker who can write to that file can add an entry that launches an arbitrary executable when the user clicks it in the Run menu.
“The injected commands appear with legitimate-looking names in the Run menu, making them appear as normal user-created shortcuts,” the advisory said. “This creates a viable persistence mechanism, as the injected commands survive reboots.”
The proof of concept Ho published shows an injected entry named “System Update Check” that launches Windows Calculator. Italian researcher Michele Piccinni reported the flaw.
A second path through the command-line interpreter
The second code execution bug, CVE-2026-48778, targets a different file. Notepad++ stores the path to its command-line interpreter in a file called config.xml and accepts whatever value it finds there as the program to launch when the user opens a folder in cmd, a separate advisory said. The interpreter path is stored “without any validation, whitelist, or digital signature check,” the advisory said. An attacker who edits config.xml can substitute any executable for the real Windows command prompt. Piccinni reported this one as well.
Neither flaw lets an attacker reach the XML files on their own, the advisories said. Both assume the attacker already has the ability to write to the user’s AppData directory or can trick the user into running Notepad++ against a poisoned settings folder, whether through local malware, a malicious Windows shortcut, cloud-synced settings, or a social-engineered archive extraction.
The third patched flaw, CVE-2026-48770, follows the same theme of unchecked input but stops short of code execution. A local process in the same Windows session can send the editor a malformed inter-process message that reliably crashes it, the advisory added. The bug carries a CVSS score of 5.0.
A question mark over MSI patch delivery
Notepad++ users can download the patched 8.9.6.1 binaries from the project’s download page, which offers both the EXE installer and an MSI installer for enterprise IT deployment that Ho added in November 2025.
The MSI followed sustained enterprise demand that intensified after a Chinese state-sponsored group hijacked the editor’s update infrastructure for six months in 2025 and after Ho hardened the update mechanism in February with cryptographic integrity checks.
The advisories recommended that users monitor the AppData folder on machines running Notepad++ for unexpected changes to shortcuts.xml and config.xml. The persistence of both flaws leaves no trace at the installation directory and no change to the Notepad++ binary itself, the advisories said, which means endpoint tools that look only at executables will miss it. Ho published no indicators of compromise.
View the full article
CSOonline

The Gentlemen are coming for your files, and then your network

Ransomware operators have spent years refining the art of locking files. Now, some are working harder to get those lockers to every reachable system first.
Microsoft’s recent warning of the Gentlemen ransomware revealed its operators using a self-propagating Go-based encryptor capable of moving laterally through compromised environments and deploying itself across additional systems.
“Modern ransomware is no longer just about encrypting files,” said Paul Reid, vice president of Adversary Research at AttackIQ. “The bigger risk is how quickly a single compromised machine can become a broader business disruption.”
In a technical breakdown of its operations, Microsoft said the Gentlemen Ransomware was first observed in mid-2025 and remains highly active through 2026, impacting organizations across education, transportation, healthcare, and financial industries in North America, South America, Europe, Africa, and Asia.
Gentlemen began as a “closed ransomware,” turned into a ransomware-as-a-service (RaaS) offering in September 2025, and eventually partnered up with BreachForums to pick up affiliates, including pen-testers and initial access brokers, from the popular cybercriminal marketplace.
Built to move before it encrypts
Microsoft’s analysis specifically focused on the ransomware’s ability to propagate through a network without relying entirely on manual operator intervention.
The encryptor, written in Go, includes functionality designed to identify additional systems, authenticate using harvested credentials, and copy itself to remote machines over Server Message Block (SMB). Once deployed, it can execute remotely and continue spreading, creating a chain infection inside compromised environments.
According to Microsoft, the malware leverages legitimate administrative tools and Windows functionality to facilitate movement while reducing the need for attackers to remain actively engaged through the operation.
“The ransomware operator can control The Gentlemen encryptor through command-line arguments,” Microsoft said. “A password is required for execution, and optional arguments allow the operator to specify encryption scope, speed, lateral movement, and post-encryption behaviors.”
One of the command line arguments,“–full,” launches separate processes to encrypt local drives with SYSTEM privileges and network shares visible to the user, to maximize encryption coverage once the machine is compromised. Additionally, a “–spread” command is used for lateral propagation.
“Defenders should treat The Gentlemen as an attack-path problem, not just a patching or detection problem,” Reid said. “The priority is to understand where the ransomware could move, which controls would detect, contain, or disrupt it, and where gaps still exist before an incident occurs.”
Gentlemen performs a “password check” to validate the use of its RaaS by the affiliates, and blocks its usage from unwanted binary recovery or interception. “Before executing its primary functionality, the malware validates the –password argument against a hardcoded value embedded within the binary,” Microsoft noted. “For the sample analyzed in this blog, the expected password is ‘9VoAvR7G’.”
Detection windows are shrinking
Microsoft’s analysis highlights the defensive challenges posed by self-propagating ransomware. Once execution begins, the time available to detect, investigate, and contain malicious activity can shrink considerably as the malware spreads to additional systems.
“This is not the kind of threat where an organization can wait for a help desk ticket or a locked screen to realize something is wrong,” said John Joyner, Senior Director of Technology at Corsica Technologies. “Malware can move quickly through a network once it gets a foothold, which makes early detection the difference between a contained incident and a business-wide disruption.”
Microsoft emphasized the importance of monitoring lateral movement activity, credential abuse, remote execution attempts, and other behaviors associated with Gentlemen’s propagation rather than focusing solely on encryption events.
Additionally, it shared a list of indicators of compromise (IOCs) to support detection efforts. For those who don’t catch it on time, the ransomware leaves a note. “Your network is locked by the Gentlemen,” a desktop wallpaper reads on the victim’s machines.
View the full article
CSOonline

Cybersecurity trends in SEC filings

In 2023, the Securities and Exchange Commission (SEC) required public companies to include a new section in their 10-K annual filings that is devoted to cybersecurity. This section is meant to address “cybersecurity risk management, strategy, governance and incidents.” I got curious as to what senior cybersecurity executives are conveying about their companies in these reports. I turned this into a research project that also gives me a reason to test out some AI techniques as well.
The article is broken into two sections: My findings regarding Section 1.C for the top 200 companies in the S&P, and the second being my methods used to include some AI tech.
10-K Section 1.C
Some really great analysis of Section 1.C has already been done to include a Harvard Law School study, a PWC study and an International Journal of Accounting Information Systems paper. These were great reads, but both were done over a year ago with the first batch of filings. Also, with the Harvard Law study, they only looked at the top 100 companies. I wanted to see if I could reproduce some of the analysis using this year’s filings, as well as ask some of my own questions, like whether there are any major changes between 2024 and 2025.
Companies are required to disclose governance regarding cybersecurity risks. Key requirements include describing board oversight of cyber risks, the committee responsible and management’s role in assessing and managing material cybersecurity threats. Years of experience are often included.
Similar to the Harvard study, I’ll look at who holds the senior cybersecurity role and their level of experience, who they report to, what part of the board oversees cybersecurity and standards that they are using. Not every company included all these pieces of information, but the bulk of them did. I’ll also look at overall trends between 2024 and 2025.
CISO role top for cybersecurity
The chief information security officer (CISO) continues to be the principal position responsible for cybersecurity, with over 70% of companies reporting CISO as the role responsible for cybersecurity.  Numbers for CISO slightly increased from 2024 to 2025, going from 137 to 142.  A distant second and third are CIO and CSO.  The average years of experience for the role is about 23 years (standard deviation 6 years, 140 companies reported).
CIO remains top senior in a varied field
Chief information officer remains the top person that the cybersecurity official reports to and remained stable between 2024 and 2025 (~49 vs ~48).  This is consistent with surveys and other reporting that CIO is the most frequent.  I agree with another CSO article that having the position under the CIO is sub-optimal and both inserts conflicts of interest as well as downplays the importance of cybersecurity at the enterprise level.  Not saying it can’t work, but there are likely better arrangements.  No clear alternative has appeared in either 2024 or 2025 data (see the chart below), and the small relative numbers indicate there is a lot of variety in who the CISO reports to.  The CEO, CFO and CTO were other common reporting positions, but none were a clear second.  It is also worth noting that for over 50 companies, it wasn’t clear from the 10-K write-ups who the reporting position was.
Derek Dye
Board oversight
Within the Company’s board, the Audit Committee is by far the most common group responsible for cybersecurity, representing 60% of companies.  This jumps to about 70% (138 companies), If you include all the variations of Audit to include Audit & Risk, Audit & Finance, etc.  Overall, audit numbers remained steady between 2024 and 2025.  Distant second and third were the Risk Committee and Board of Directors broadly.
NIST CSF for the win
National Institute of Standards and Technology (NIST) Cybersecurity Framework (CSF) is the most referenced cybersecurity standard, increasing between 2024 and 2025(113 vs 118). The most common other standard being ISO 27001, which also grew between 2024 and 2025 (49 vs 55).  Interestingly, System and Organization Controls (SOC) was only mentioned by 17 companies.  I find this seemingly low, given the importance of SOC reporting in large public sector companies.
Overall trends
Other interesting observations were what companies listed as their broad efforts as well as disclosures of incidents.
Third-party and supply chain risk management. Acknowledging that external partners and suppliers represent a massive attack vector, multiple companies have instituted rigorous third-party risk management (TPRM) programs. These programs mandate pre-engagement security assessments, continuous monitoring and contractual requirements for vendors to maintain security standards and report breaches promptly.  Third-party cybersecurity programs are indispensable in an increasingly interconnected economy and increasing reliance on external tools and services for company processes. Proactive testing and incident preparedness. Companies are moving past passive defense into proactive and simulated testing. This includes regular penetration testing, vulnerability scanning and engaging independent external auditors or consultants to assess program maturity and test controls. Furthermore, practically all companies maintain formal Incident Response Plans (IRPs) and conduct regular “tabletop exercises” to simulate cyberattacks, ensuring that management, legal and operational teams are prepared to respond to and recover from real-world crises. The devil is in the details on this one. The 10-K is not meant as a detailed technical rundown of company methods, so while it’s good to see, it’s mostly boilerplate language. Human-centric security defenses. Recognizing that human error is a primary vulnerability, mandatory, enterprise-wide cybersecurity awareness training is a standard requirement. These training programs are frequently supplemented with regular, simulated phishing campaigns to test employee vigilance and provide immediate, targeted feedback or remedial training.  This training will need to adapt to the growing sophistication of AI-enabled deep fakes. Consistent disclosure of “No Material Impact” despite ongoing threats. A ubiquitous trend across the filings is the acknowledgment that while the companies face continuous, sophisticated and evolving cyberattacks, they have not experienced any incidents that have had a material adverse effect on their business strategy, results of operations or financial condition to date. I find this interesting, especially with the Critical Infrastructure//telecoms coming under repeated VOLT/SALT TYPHOON compromises as well as other attacks. Many companies also disclose that they rely on cyber liability insurance to mitigate financial exposure, though they frequently note it may not cover all potential losses.  I’ll be doing further research here as there are likely more interesting findings between material impacts, news reporting and formal disclosures. Artificial intelligence. AI was cited by over 50 companies and is increasingly referenced as a double-edged sword for cybersecurity.  Companies are leveraging AI and machine learning to automate threat detection and sort through vast amounts of security data. However, several acknowledged that AI empowers threat actors to execute more sophisticated, high-velocity attacks (e.g. deepfakes, advanced phishing).  A further seven companies mentioned the concern of AI and intellectual property disclosures with Prudential and Capital One having the most explicit language on this risk. Part 2: Data gathering and analysis
This was a very iterative process that increased in complexity as I went through the process and also due to the increased need for accuracy.  I used several coding methods and AI tools to do this analysis. At first, I tried to use the big models to do all the work for me, but that quickly failed when they didn’t want to do that level of work! It also became apparent that getting the 10-K filings would take more work than just asking an AI agent.
Enter some vibe coding. I was raised on C, Java and BASH scripting and have avoided using Python until now. Nothing against Python, I just haven’t needed to, and laziness with going with what you already know has won out before. So, this proved a nice additional challenge. Using the datamule Python module and some vibing, I managed to download all the recent 10-Ks for the top 200 companies onto my local machine. From there, I extracted the 1.C sections into a separate file using another Python script. This caused a bit of an issue as there were some differences (~5%) in filings that used a different format, or the cybersecurity write-up was in a different section of the 10-K.  About 15 companies put them in the Risk section or elsewhere.  I used a second Python script that leveraged the command-line version of Gemini (gemini-cli) to pull this information out.
I then created a database in postgres that would store some of the key findings and allow for some further analysis. To get the data into there, I created a Python script that would run each of the 1.C files through Gemini and Claude using Python API calls.
The use of Gemini API and Anthropic API was the new part that I really wanted to test out, and it proved very interesting. LLMs really shine for condensing and summarizing large texts for meaning. The alternative would be very complex and manually written regular expressions. Using Gemini API and Anthropic API, it took the below prompt and produced a string that I could then plug into the SQL command. Very cool seeing this work. (**Note: I was also thinking of how to do prompt injection, data poisoning and the like with this, but the dataset was small and controlled and this isn’t production code!).
Derek Dye
As a verification step, I then wrote another script that found all database entry differences between Gemini and Claude answers and ran the original 1.C section through Gemini again and told it to pick which answer was better.  This changed about 10-30% of the entries, depending on the field.  Additional analysis was done in Google Sheets and Google NotebookLM.
With this, I created a basic AI-enabled workflow. It wasn’t agentic, but that would be interesting to create an automated version of this.  This project showed some of the productivity potential of AI by allowing me to do very detailed research in about 15-20 hours of work, which would have taken at least twice as long by hand.  It also highlighted the continued issue with accuracy where accuracy is needed.  The bulk of the 15-20 hours was spent doing verification and refinement to make sure the AI answers were correct.
The total cost in tokens for development, debugging, and running was around $15, not expensive, but not something I’d likely develop for every project I have. The bulk of that cost came with the refinement and the addition of additional verification checks to ensure the data was correct.  Next projects might try to do this on my local computer using a local LLM like llama3 using ollama or maybe an agent that allows queries to the dataset this project created.
GEMINI_PROMPT = """ Analyze this SEC 10-K document and extract the following cybersecurity information. Return the response strictly as a JSON object with these exact keys: { "senior_cyber": "Name or title of the senior person responsible for cybersecurity. Provide a one word response either CISO, CTO, CSO, CIO, or position title.", "report_to": "Title or name of who the senior cybersecurity person reports to. one word response either CEO, CTO, CSO, CIO, position title, or unknown. ", "board": "The board committee overseeing cybersecurity Provide a 1-3 word answer. ", "standards": "The cybersecurity standards/frameworks used use provide 5-7 word answer. If unknown, state unknown. use acronyms if available (e.g., NIST, NIST CSF, NIST CSF 2.0, ISO 27001)", "years_of_experience": integer representing years of experience (use 0 if unknown) } """ MODEL_ID = 'gemini-2.5-flash' —---- # 6. Update PostgreSQL upsert_query = """ INSERT INTO company_cyber_filings_v1_4 (ticker, filing_date, senior_cyber, reports_to, board, st andards, years_of_experience) VALUES (%s, %s, %s,%s,%s,%s,%s) ON CONFLICT (ticker, filing_date) DO UPDATE SET senior_cyber = EXCLUDED.senior_cyber, reports_to = EXCLUDED.reports_to, board = EXCLUDED.board, standards = EXCLUDED.standards, years_of_experience = EXCLUDED.years_of_experience; """ # 4. Upload file formatted_date = f"{year}-{month}-{day}" # Formatted for standard SQL DATE cursor.execute(upsert_query, ( prefix, formatted_date, gemini_data.get("senior_cyber"), gemini_data.get("report_to"), gemini_data.get("board"), gemini_data.get("standards"), gemini_data.get("years_of_experience") )) cursor.execute(upsert_query, (prefix,f"{year}{month}{day}")) conn.commit() print(f" -> Saved to database.") This article is published as part of the Foundry Expert Contributor Network.
Want to join?
View the full article
CSOonline

GDPR set the tone for regulatory action — and the AI fine pushback to come

Big tech firms continue to push back against fines levied for alleged violations of European data protection law, in what could be a harbinger for AI regulations to come.
While lawyers and experts quizzed by CSO broadly argue that big tech firms contesting data protection rules isn’t a particular cause for concern, the more widespread introduction of AI technologies is a far greater data protection challenge on the horizon.
The EU’s General Data Protection Regulation (GDPR) came into force eight years ago this week. Over those eight years, European regulators announced an estimated €7.1 billion in GDPR fines but nearly 40%, around €2.8 billion, has either already been annulled or is under active legal challenge, according to analysis by insurance brokerage Alliance Risk.
Fines that have already been annulled include one against Amazon at €746 million (Luxembourg, March 2026) and another versus OpenAI at €15 million (Italy, March 2026). Those under active appeal include three fines against Meta (€1.2 billion, €265 million, and €91 million) and one against TikTok (€530 million).
Alliance Risk used CMS Law GDPR Enforcement Tracker as its primary source for information on GDPR enforcement, cross-referenced against IAPP enforcement data and trackers from Kiteworks and UniConsent. Data on annulments came from reported court decisions.
GDPR established a benchmark for breach notification
According to Alliance Risk, GDPR successfully laid the foundation for data protection law globally — particularly by first establishing the 72-hour breach notification standard.
This three-day notification rule is law in six jurisdictions — EU, UK, Thailand, Kenya, Nigeria, and South Korea — and influential elsewhere. For example, the US CIRCIA rule for critical infrastructure, which is pending final rule publication this month, is due to apply the 72-hour standard.
By comparison, HIPAA gives US healthcare organisations 60 days as a breach notification deadline. The SEC gives public companies four business days but only after they’ve internally determined a breach is “material,” which adds its own delay.
Although the breach notification regulations established by GDPR have been a success, issues with the enforcement of rules remain.
“The framework has structural weaknesses that large companies have learned to exploit in court, and nearly 40% of announced fines reflect that,” according to Alliance Risk.
The EU’s AI Act reaches full application in August, and the European Commission is already proposing to reform GDPR through the Digital Omnibus. “The framework is being rewritten while it’s still being tested,” Alliance Risk concludes.
“The fact that around 40% of GDPR fines by value are under challenge isn’t necessarily a sign the system is broken,” Nick Phillips, an intellectual property lawyer at Edwin Coe LLP tells CSO. “Eight years in, the bigger fines were always going to end up in court, and the rulings that come out of those appeals are starting to give in-house teams something they’ve never really had before: practical guidance on what regulators can and can’t defend.”
Phillips argues that achieving compliance with GDPR has improved enterprise security maturity because of the 72-hour breach notification rule coupled with the obligation to record all breaches and to notify data subjects combined with the need to improve security controls even more than the threat of a fine for non-compliance.
“That breach notification regime has arguably been the single biggest factor in forcing organisations to put proper incident response in place, get forensics providers on retainer, and start reporting breaches up to the board,” Phillips says. “A lot of that simply wasn’t happening before 2018, and it’s the part of GDPR that’s done the most work.”
Marco Eggerling, LL.M, security and trust officer EMEA and Asia, at robotic process automation vendor UiPath, says it would be a “mistake to read these annulments as courts clearing big tech.”
“In the Amazon case, the Luxembourg court upheld the substance of the violations and sent the matter back to the regulator,” Eggerling notes. “The fine fell because the authority skipped required steps, not because the conduct was found lawful.”
Eggerling adds: “The lesson for regulators is to build procedurally bulletproof decisions. The lesson for companies is that the underlying obligations have not moved an inch.”
Even within the EU there is a disparity in how regulations are understood and applied, making cross-border decisions about data and AI challenging.
“A lot of organisations lean towards the ‘lowest common denominator’ and adhere to the strictest governance and more conservative approaches in order to avoid the wrath of regulators,” says Caroline Carruthers, CEO and founder of global data consultancy Carruthers and Jackson.
The UK and EU apply stricter regulations than the US or China, so many organisations adhere to the stricter rules wherever they operate.
Due to their size and nature, “big tech” organisations tend to have a heightened appetite for risk and a desire to push the boundaries of regulations — and often a different relationship with the general public, whose data is the business model. “They have a vested interest in deregulation and so will naturally be the most likely to contest enforcement,” Carruthers notes.
Data regulations need to evolve with the advent of AI
For most organisations, the enforcement of GDPR has gotten to a place where it is broadly fit-for-purpose, according to Carruthers.
“When GDPR was first introduced, the guidance was unclear and inconsistent,” Carruthers explains. “It felt legally robust, but a lot of the data practitioners struggled to make it work. Even now, some businesses tell us that they are ‘paralysed’ a little by GDPR. They are highly fearful of data and the associated regulation, to the extent that they are unable to maximise — or even touch on — the potential power of data.”
However, as AI and data regulation evolves, there’s a need to account for how these tools are now being used.
The concern is that history may repeat itself as regulation looks to keep pace with technological change. “There is a risk that organisations get stuck in a mid-maturity plateau in which innovation is halted by complex and inconsistent interpretations of regulations,” Carruthers warns.
View the full article
CSOonline

Kimsuky Deploys HTTPSpy, Expands Arsenal with HelloDoor and VS Code Tunnels

The North Korean state-sponsored threat actor known as Kimsuky (aka Velvet Chollima) has been attributed to a fresh set of cyber attacks targeting South Korean military and corporate entities through March and April 2026. "Kimsuky employed a range of tailored social engineering tactics, such as spoofing security software installation pages and crafting a fake Webex meeting page that leveragedView the full article
Hacker News

IBM and Red Hat want to become the ‘security clearinghouse’ for open source applications in the enterprise

Open source code is everywhere in the enterprise; it’s estimated that upwards of 90% of Fortune 500 companies have it in their software supply chains. But open source code is notoriously rife with vulnerabilities, and identifying and patching those bugs can be an endless battle for security teams.
IBM and Red Hat are betting that a new initiative, Project Lightwell, can help accelerate this process.
Announced today, the project will commit $5 billion and 20,000 IBM and Red Hat engineers to build a new ‘enterprise clearinghouse’ to accelerate discovery and remediation of vulnerabilities in open source software. The companies say the clearinghouse will serve as an AI-powered  “security coordination layer,” giving enterprises the ability to integrate patches directly into their existing software supply chains.
Now in the design phase with a group of 11 financial partners, Project Lightwell will eventually be offered as a commercial subscription.
“The advancement in AI tools has broken the patching map, which is the ability to discover vulnerabilities in software without losing the speed of remediation,” Ashesh Badani, Red Hat SVP and CPO, told CSOonline. “Everyone’s running open source software, and the challenge is not being able to fix vulnerabilities quickly enough.”
Closing the remediation gap
Open source security issues have been well documented: Almost 50,000 common vulnerabilities and exposures (CVEs) were published in 2025, and Anthropic’s Project Glasswing, powered by its Mythos Preview model, found roughly 3,900 previously undiscovered high or critical severity vulnerabilities in open source software shortly after launch.
IBM is considered one of the broadest commercial open source ecosystems, using more than 62,000 packages and operating across Linux, Kubernetes, Kafka, Terraform, Java and other platforms, and providing lifecycle management, validation, and patching for elements within those environments.
The company says Project Lightwell will now apply those same engineering principles to broader AI frameworks, independent libraries, language toolchains, and data streaming platforms, to deliver validated fixes to open-source code already in use in enterprise environments. This can support remediation without disruption of stability, certification, or compliance.
No upgrades or access to source code are required; Project Lightwell will backport fixes to exact dependency versions that have already been tested and deployed. It operates on fundamental configuration manifests like pom.xml so code remains in controlled enterprise environments when patched artifacts are rolled out. Initial focus will be on Java/Maven, but the project will eventually expand to PyPI, npm, Go, and others.
Enterprises will have the ability to share sensitive vulnerabilities under embargo through a “secure intermediary model” and receive validated patches spanning Red Hat platforms and independent community code. They will also be able to deliver fixes across dependency chains; report and address issues across active production environments; and share fixes upstream so the wider open-source community can incorporate them.
“We want to make sure that whatever fixes we provide to the enterprises through the clearinghouse also find their way back into the open source community that developed [the code],” Badani explained. For instance, if a piece of Python code was patched, the fix should be quickly delivered back to the Python community. With Project Lightwell, that process can be achieved through a “secure map.”
Using advanced AI, and working with leading open source contributors, IBM and Red Hat engineers will focus on connecting upstream and downstream environments so fixes are enterprise-ready. They will also develop patches and perform “high volume” vulnerability review and triage, and dependency hardening.
The network of 20,000 engineers will come from IBM’s and Red Hat’s existing pools of talent, and the companies will augment those teams as needed, Badani explained. The companies will take advantage of foundation models coming out of frontier labs, as well as their own internally-built AI tools and frameworks. The $5 billion will be used to equip teams with AI tools and build out internal operational infrastructure.
Early Project Lightwell adopters include Bank of America, BNY, Citi, Goldman Sachs, JPMorganChase, Mastercard, Morgan Stanley, Royal Bank of Canada, State Street, Visa, and Wells Fargo. Following the initial design period, IBM and Red Hat will phase more customers onto Project Lightwell via a subscription model.
A call to action?
This type of initiative is “desperately needed” if enterprise is to save open source, noted David Shipley of Beauceron Security.
The days of trillions in wealth depending on volunteers “ended violently” with Mythos, he noted, and the bill has ultimately come due for open source. Enterprises will need to pay up, or lose it.
“If we don’t find a way to invest in open source, which will close a long-standing equity issue, the alternative is everyone building their own bespoke code using AI,” Shipley said. That would be “massively wasteful” from a compute and environmental perspective.
“I hope this drives others to act,” he said.
Keeping humans in the loop for an ongoing battle
Badani emphasized that, while AI is great at discovering security issues in open-source code, the patching process can still be cumbersome. Fixes have to be sent upstream, distributed to the open source community, then flow back to customers and users.
“Finding the bug is one thing,” said Badani. “The other is all the steps that it takes to actually go and remediate it. That extra amount of time is the gap that we’re trying to help close.”
Underscoring the severity of the problem, IBM and Red Hat have already had an “onslaught of incoming requests” since Project Lightwell was announced.
“This isn’t going to stop any time soon,” Badani said. “Even if we were to very successfully solve the initial set of challenges that come to us, this will be something that companies are going to need on an ongoing or recurring basis.”
And, while the narrative has focused on cutting human engineers in favor of AI, Project Lightwell is focused on the opposite: “We can address [the problem] with a mixture of AI tools and human knowledge and expertise,” Badani said. “Coupling the two gives you a better outcome than just using one or the other.”
This article originally appeared on InfoWorld.

View the full article
CSOonline

Lack of response to critical vulnerability in Gogs is a reminder of the limits of open source projects

A newly discovered and so far unpatched critical vulnerability in the open source Gogs Git service not only demands immediate action from developers to secure their code, it also puts a spotlight on the potential issues in using self-hosted code platforms from small maintainers.
The hole is a critical argument injection vulnerability, discovered by a researcher at Rapid7, that allows any authenticated user to remotely execute code on a Gogs server by creating a pull request with a malicious branch name during a merge operation.
Rapid7 published an analysis of the vulnerability today, after the maintainer of Gogs did not respond to a request for status updates or to an offer to defer disclosure after it first reported the hole over two months ago.
“This is a serious vulnerability in software that isn’t commonly exposed to the public internet,”  Ryan Emmons, staff security researcher at Rapid7, said in an email.
“Gogs is typically used in an internal capacity; the most likely threat model is an attacker that has already gained access to an internal network environment exploiting the vulnerability to gain read/write access to source code repositories on the Gogs server. An attacker might leverage this access to silently tamper with source code and exfiltrate sensitive information, such as user password hashes and proprietary software.”
Rapid defensive action required
David Shipley, head of security awareness provider Beauceron Security, said both the Gogs maintainer and developers must take defensive action fast, because with the publication of a vulnerability “any attackers that didn’t know about this are going to be on it viciously.”
The fact that it has been left unpatched for months as of Thursday afternoon is another reason why CSOs and developers prefer GitHub, he added. With any open source project, there are worries about if or when a patch will be issued.
“The exploit requires no admin privileges and no interaction with other users,” Rapid7 said in its report. “An attacker operates entirely within their own account. Since Gogs ships with open registration enabled by default (DISABLE_REGISTRATION = false) and no limit on repository creation (MAX_CREATION_LIMIT = -1), an unauthenticated attacker can simply create an account and repository on any default-configured instance. Any registered user who creates a repo is automatically its owner. From there, enabling rebase merging is a single toggle in settings, and the entire exploit chain can be operated without interaction from any other user.”
In addition, any user with write access to a repository where rebase is already enabled can exploit it directly. On instances where repository creation is restricted, an attacker still only needs write access to any repository that has (or can have) rebase merging enabled.
If exploited, the vulnerability could not only lead to a Gogs server compromise, but from there it could turn into to a cross-tenant data breach, credential theft, lateral movement across an IT network, and software supply chain attacks through the code that is being developed on the compromised Gogs platform.
Until a patch is released, developers and CSOs in organizations with the platform in use should strictly enforce restricted network access to Gogs, Emmons said, and ensure that only those who need access can use the application. Furthermore, if user self-registration is not already disabled, it should be. Only administrators should be able to create new user accounts.
Rapid7 describes Gogs as a lightweight, self-hosted Git service written in Go that can run on any platform supported by the Go toolchain, including Linux, macOS, and Windows, as well as on ARM-based systems. It’s one of the more popular self-hosted alternatives to Microsoft-owned GitHub, says Rapid7, and is commonly deployed by companies, universities, and open-source projects.
Other self-hosted Git services for developers include GitLab Community Edition, Gitea, Forgejo (a fork of Gitea), and Atlassian’s Bitbucket Data Center.
Gogs pros and cons
In a blog earlier this month, Open Source Alternatives, which describes itself as a curated directory of self-hosted tools that replace paid software, noted that developers may chose to self-host a git server to avoid GitHub outages, arguing, “your repositories stay online when GitHub goes down, your GitHub Actions minutes bill disappears and your source code never leaves your own server”.
Emmons said Gogs is popular because it’s a lightweight and self-contained Git solution. It’s easy to deploy and run, he said, unlike many other Git servers that require heavy operational overhead and IT management. It’s also self-hosted on-prem software, which he said is ideal for teams that don’t, or cannot, for one reason or another, store source code in the cloud.
The main pro, Emmons said, is that Gogs is an appealing solution from an operational simplicity perspective. It works well for what it does, and it doesn’t take much management effort to keep it working. But, he added, “a major con is what we saw with this disclosure; Gogs is open-source software maintained by kind people in their free time, and the developers behind it don’t have the support of a major corporate information security team. That means security issues can sometimes present in ways that they typically wouldn’t for a well-funded enterprise product.”
This article originally appeared on InfoWorld.
View the full article
CSOonline

Anthropic Launches Claude Opus 4.8 With Gains in Coding and Honesty

Anthropic today announced the launch of its latest AI model, Claude Opus 4.8. Anthropic claims the model is a "more effective collaborator" with improvements in agentic coding, multidisciplinary reasoning, agentic computer use, knowledge work, and agentic financial analysis.


Testers have found Opus 4.8 to be "more reliable and sharper in its judgement" when doing agentic tasks, and the model also made gains in honesty.

Alignment assessments suggest the model hits new highs on measures of prosocial traits like supporting user autonomy and acting in the user's best interest. Rates of misaligned behavior like deception are lower than Opus 4.7 and similar to the Claude Mythos Preview.

Anthropic benchmarks indicate Opus 4.8 scored a 69.2% on SWE-Bench Pro, outperforming GPT–5.5 and Gemini 3.1 Pro on the test and several other benchmarks, though GPT–5.5 leads on the terminal-coding benchmark.

Opus 4.8's fast mode also runs at 2.5x the speed, and it is now three times cheaper than prior models.

Along with Opus 4.8, Anthropic is adding new features to its product lineup.

Dynamic workflows (research preview) - Claude can complete bigger tasks in Claude Code. It is able to plan work and run hundreds of parallel subagents in a single session. It is able to complete codebase-scale migrations across hundreds of thousands of lines of code. The feature is available for Claude Code for Enterprise, Team, and Max plans.
Effort control - In Claude.ai and Cowork, users can choose how much effort Claude puts into a response. With a lower setting, Claude will respond faster and use up rate limits more slowly. Opus 4.8 defaults to high effort, which Anthropic says is the best balance of quality and user experience.
Messages API - The Messages API accepts system entries inside the messages array, so developers can update Claude's instructions mid-task.

Claude Opus 4.8 is available everywhere today. Pricing for regular use has not changed compared to Opus 4.7.

Anthropic is working on models that have the same capabilities as Opus 4.8 at a lower cost, and a new class of model that's even more intelligent than Opus. Anthropic says it has been developing safeguards for the Claude Mythos model it is testing with a small number of organizations, and it expects to be able to bring Mythos-class models to all customers "in the coming weeks."Tag: Anthropic
This article, "Anthropic Launches Claude Opus 4.8 With Gains in Coding and Honesty" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Seeds First Public Betas of iOS 26.6, macOS Tahoe 26.6 and More

Apple today provided public beta testers with the first betas of iOS 26.6, iPadOS 26.6, macOS Tahoe 26.6, watchOS 26.6, and tvOS 26.6, with the software coming two days after Apple seeded the betas to developers.


After signing up to beta test the software updates on Apple's beta site, public beta testers can download the new software using the Software Update section in the Settings app on each device.

iOS 26.6 has a feature that will let you know when you have blocked too many contacts, but the limit is in the thousands so most users may not ever see the messaging.

No other major new features have been found in any of the software updates, with Apple likely focusing on bug fixes and security improvements. We're nearing the end of the "26" software cycle, with Apple planning to unveil iOS 27, iPadOS 27, macOS 27, and more in 11 days.Related Roundups: iOS 26, iPadOS 26, macOS TahoeRelated Forums: iOS 26, macOS Tahoe
This article, "Apple Seeds First Public Betas of iOS 26.6, macOS Tahoe 26.6 and More" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Hold Off on Purchases, Students: Apple's Back to School Sale Is Coming Soon

Apple has started preparing for its annual Back to School promotion in the U.S. and Canada, according to backend Apple code seen by MacRumors.


Apple holds a Back to School sale each June, providing teachers and students with free accessories or gift cards when purchasing a Mac or an iPad. The added bonus is in addition to Apple's educational discount.

Back to School sales typically start around the middle of the month, but we don't yet know the start dates for this year's sale. The sale continues through the end of September, so students do not need to rush to prepare to return to school.

Apple's Back to School promotions have varied. Last year, Apple offered eligible students, educators, and parents accessories worth up to $199 when purchasing a Mac or an ‌iPad‌. Free options included the AirPods 4 with Active Noise Cancellation, Magic Mouse, Magic Trackpad, Magic Keyboard with Touch ID, Apple Pencil Pro, and more.

Looking back at Apple's past Back to School dates gives us an idea of when this year's sale will take place.

2021 - June 17 to September 27, free AirPods
2022 - June 24 to September 26, gift card up to $150 U.S.
2023 - June 5 (U.S.)/June 26 (Canada) to October 2, gift card up to $150 U.S.
2024 - June 20 to September 30, gift card up to $150 U.S.
2025 - June 17 to September 30, free accessory worth up to $199

The Back to School sale typically takes place around WWDC, and in three of the last five years, it has started 8 to 10 days after the WWDC keynote. With the 2026 keynote set to take place on June 8, we're likely to see the sale start the week of June 15.

Students who are planning to purchase a new Mac or ‌iPad‌ may want to hold off until Apple's Back to School sale begins to get an added accessory or gift card.

In past years, MacBooks, iPads, and iMacs have been eligible for a free add-on with purchase.

Apple holds the promotion through its online education store and Apple retail locations. Students attending or accepted to a higher-education institution, parents purchasing on behalf of an eligible student, and employees at K–12 schools and higher-education institutions are eligible.

Earlier this year, Apple began requiring UNiDAYS verification for educational discounts in the United States and Canada. Educational purchases now require academic status verification through an email address from an educational institution, a student or staff photo ID, or another valid educational document.

Apple actually holds two Back to School sales per year, with one aimed at students in Australia, New Zealand, and other Southern Hemisphere markets. This year's January Back to School sale saw Apple offering students free AirPods, Apple Pencils, and other accessories.

After the U.S. and Canada Back to School sales launch, Apple typically extends them to Europe, the Middle East, and Asia a few weeks later.Tag: Back to School Promotion
This article, "Hold Off on Purchases, Students: Apple's Back to School Sale Is Coming Soon" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Rivian Explains Why CarPlay Debate Will Become 'Completely Obsolete'

On the latest episode of The Verge's Decoder podcast, Rivian's software chief Wassym Bensaid explained why the EV maker still refuses to offer Apple CarPlay.


In short, Bensaid said Rivian does not want CarPlay to fully take over the software experience.

"The challenge with screen mirroring solutions is that they take over every single pixel in the car," he said.

Instead, Rivian prefers to offer its own interface with "end-to-end integration."

Bensaid added that "deep AI integration into the car" will eventually make the debate over offering CarPlay in vehicles "completely obsolete." He believes that Rivian owners will be able to access the core functionality of many apps through an AI agent, rather than needing to rely on CarPlay or Android Auto.

"I really believe that the way you interact with apps which are mono-threaded, single buttons, single icons, a lot of that will be now completely reshaped into a world where it will become an agentic integration that presents itself into a wholesome user experience to the user," Bensaid explained. However, he admitted that an AI agent becoming an "alternative solution" to CarPlay apps will take time to materialize.

Already, Rivian owners are increasingly less interested in CarPlay, according to Bensaid. He said that Rivian's internal statistics showed that more than 70% of customers wanted CarPlay when their vehicles first launched around five years ago, but he said a recent survey showed that figure is apparently less than 25% now.

"With the level of features that we have shipped to customers, with the level of end-to-end integration, with the level of convenience that we are bringing, CarPlay is, or Android Auto to that extent, is no longer the topic of discussion," he said.Related Roundup: CarPlayTag: RivianRelated Forum: HomePod, HomeKit, CarPlay, Home & Auto Technology
This article, "Rivian Explains Why CarPlay Debate Will Become 'Completely Obsolete'" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Oura Ring 5 Unveiled With Smaller Design and Blood Pressure Tracking

Oura today unveiled the Oura Ring 5, a significantly smaller smart ring with new health-monitoring capabilities including blood pressure trend detection, nighttime breathing analysis, and tools for GLP-1 medication tracking.


The Oura Ring 5 is 40% smaller than its predecessor, measuring 6.09mm wide and 2.29mm thick compared to 7.99mm and 2.88mm on the previous model. Oura says the reduction was achieved by redesigning the mechanical, electrical, optical, battery, and sensing architectures, and the company claims the new form factor more closely resembles a traditional wedding band. Despite the smaller size, Oura says battery life remains at approximately one week per charge.

The ring's sensing system has also been overhauled. A new signal architecture uses precision-engineered sensor domes for better skin contact, more powerful LEDs, and twelve signal pathways that Oura says deliver greater accuracy across more finger types and skin tones.

The headline software addition is "Health Radar," a proactive health-monitoring feature that builds on the company's existing Symptom Radar tool, which launched in 2024 to flag early signs of illness. Health Radar includes two initial capabilities: Blood Pressure Signals and Nighttime Breathing.

Blood Pressure Signals continuously monitors for patterns that may indicate cardiovascular strain, surfacing nighttime blood pressure trends. Nighttime Breathing provides a 30-day rolling view of sleep-related breathing disturbances, with guidance on when to seek further evaluation.

A new Health Records feature allows U.S. users to import diagnosed conditions, medications, lab results, and allergies directly into the app. Oura has also partnered with Counsel Health, an AI-powered virtual care platform, to let eligible users in 43 U.S. states ask health questions and connect with licensed physicians from within the app for an additional fee.

New GLP-1 Insights tools let users track medication dosing, side effects, weight changes, and biometric data in one place. A Lab Uploads feature lets users import blood biomarker results alongside Oura's biometric data.

Other new features include live workout tracking with real-time pace, distance, and heart rate via a connected device, a Brain Health Study through Oura Labs pairing cognitive assessments with biometric data, and a time-based Data Deletion tool allowing users to erase data from specific periods without affecting their full history.

An optional new Charging Case accessory provides one month of battery and supports wireless charging, priced at $99. Oura is also introducing a Locate feature to track misplaced rings and charging cases.

The Oura Ring 5 is priced at $399 in Silver and Black, with Gold, Stealth, Brushed Silver, and Deep Rose finishes priced at $499. The Oura Ring 4 in ceramic remains available at $349. An Oura Membership is required for full functionality at $5.99 per month or $69.99 annually. Global pre-orders open today, with shipping beginning June 4. Tags: Health and Fitness, Health Technologies
This article, "Oura Ring 5 Unveiled With Smaller Design and Blood Pressure Tracking" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Threat Actors Exploit Critical FortiClient EMS Flaw to Deploy Credential Stealer

Threat actors are continuing to exploit a critical, now-patched security flaw impacting FortiClient Endpoint Management Server (EMS) deployments to deliver credential-stealing malware. "The campaign abused trusted endpoint management infrastructure to deliver malware across managed endpoints," Arctic Wolf said. "Threat actors disguised the credential stealer payload as a Fortinet endpointView the full article
Hacker News

Report: Apple Plans to Make On-Device AI a Key WWDC Focus

Apple reportedly plans to use next month's Worldwide Developers Conference (WWDC) to highlight its on-device AI capabilities as a competitive advantage, leaning on 15 years of custom silicon expertise to make the case for running AI models locally rather than in the cloud.


People familiar with Apple's plans speaking to The Information say the company is expected to showcase how the chips designed for iPhones, Apple Watches, and Macs give it an edge in processing AI queries directly on devices. While cloud-based processing will remain necessary for complex queries, Apple will position local inference as a privacy-preserving, cost-saving alternative to the massive data center buildouts its rivals have pursued.

As part of its agreement with Google, Apple is apparently set to use a large version of Google's Gemini model to train a smaller, distilled version capable of running locally on Apple hardware. Apple is also said to be scouting acquisitions to help advance its model-shrinking work, with one company it has reportedly considered being Liquid AI, a Massachusetts startup focused on running AI locally on devices.

Some queries will still require cloud processing. Apple is believed to have approved the use of Nvidia's confidential compute technology within Google Cloud to handle processing of the larger Gemini-based model. The security feature encrypts data and AI models during processing, adding a modest performance cost but offering stronger privacy protections.

The arrangement represents a noticeable departure from Apple's original Apple Intelligence announcement, in which the company said all cloud-bound queries would be handled exclusively by its own Private Cloud Compute infrastructure running on Apple silicon. Apple is likely to retain the Private Cloud Compute branding despite the change, people familiar with the partnership told The Information.

There are also said to be material limits to how far Apple can push on-device processing. Google's full Gemini model runs into the trillions of parameters, and The Information claims that Apple has struggled to run it on its own Private Cloud Compute infrastructure, which uses the same Apple silicon chips found in Mac computers.

‌Apple Intelligence‌ was first announced at WWDC 2024, but the rollout has been hampered by a tepid response to initial features and a protracted delay to the more personal version of Siri. Apple is now expected to use WWDC 2026, which runs from June 8 to reframe the narrative, reintroduce the delayed features, and debut new ones. Related Roundup: WWDC 2026Tags: Apple Intelligence, The Information, WWDC 2026Related Forum: Apple, Inc and Tech Industry
This article, "Report: Apple Plans to Make On-Device AI a Key WWDC Focus" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Amazon Takes $199 Off Nearly Every M5 MacBook Air, Starting at $899.99

Earlier this week, Amazon introduced a new $199 discount on the 13-inch M5 MacBook Air, and this sale has now expanded to include nearly every model of the computer on sale at record low prices. You can get the 512GB 13-inch M5 MacBook Air for $899.99, down from $1,099.00, available in all colors.

Note: MacRumors is an affiliate partner with Amazon. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

You'll find $199 off every model of the 13-inch M5 MacBook Air on Amazon, with free delivery around early June for most models. In terms of other 13-inch models, Amazon also has the 16GB/1TB model for $1,099.99, and the 24GB/1TB model for $1,299.99. Both of these represent $199 discounts and are new all-time low prices for each model.

$199 OFF13-inch M5 MacBook Air (512GB) for $899.99
$199 OFF13-inch M5 MacBook Air (16GB/1TB) for $1,099.99
$199 OFF13-inch M5 MacBook Air (24GB/1TB) for $1,299.99

Regarding the 15-inch models, you'll also find $199 off the larger M5 MacBook Air, with multiple color options on sale for each configuration. Prices start at $1,099.99 for the 512GB model, down from $1,299.00, and rise to $1,499.99 for the 24GB/1TB model, down from $1,699.00. As of writing, we aren't tracking any discounts on the 16GB/1TB model of the 15-inch M5 MacBook Air.

$199 OFF15-inch M5 MacBook Air (512GB) for $1,099.99
$199 OFF15-inch M5 MacBook Air (24GB/1TB) for $1,499.99

If you're on the hunt for more discounts, be sure to visit our Apple Deals roundup where we recap the best Apple-related bargains of the past week.



Deals Newsletter

Interested in hearing more about the best deals you can find in 2026? Sign up for our Deals Newsletter and we'll keep you updated so you don't miss the biggest deals of the season!




Related Roundup: Apple Deals
This article, "Amazon Takes $199 Off Nearly Every M5 MacBook Air, Starting at $899.99" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

First Look at AI Camera and Photos Features in iOS 27 Revealed

Apple is planning sweeping AI-driven upgrades to its Camera and Photos apps in iOS 27, Bloomberg's Mark Gurman reports.

Image via Bloomberg.

The report offers a first look at the appearance of several major ‌iOS 27‌ features that Apple plans to announce at its Worldwide Developers Conference on June 8. The images are based on information viewed by Bloomberg and people said to be familiar with Apple's plans.

The Camera app is set to gain a dedicated Siri mode, positioned alongside existing options like Photo and Video. Gurman reports the feature would replace the current Visual Intelligence experience, allowing users to photograph objects and have them analyzed by a third-party AI agent or run through a Google reverse image search. By elevating the capability directly into the Camera app rather than restricting it to the Camera Control button, Apple reportedly aims to increase adoption and help acclimate users to visual AI ahead of future products, including smart glasses and camera-equipped AirPods.

The Camera app is also said to be getting a new "Add Widgets" panel that makes the interface more customizable. The top row of shortcuts currently displayed across capture modes would become replaceable, letting users prioritize more professional controls such as depth adjustments, or surface tools like timers and Night mode more prominently in the interface. Gurman says the changes are aimed at making Apple's camera software more appealing to advanced photographers.

Image via Bloomberg.

The ‌Photos‌ app is set to receive new Apple Intelligence tools called "Reframe" and "Extend." Reframe would allow users to change the perspective of a photo, while Extend uses AI to generate additional portions of an image, such as filling in the lower half of a building that was cut off in the original shot.

Apple is also said to be testing natural language prompt-based photo editing, which would let users request specific edits by voice or text, such as cropping or adjusting colors. Gurman notes that this specific feature may not arrive in the first version of ‌iOS 27‌.

Elsewhere in ‌iOS 27‌, the Shortcuts app is said to be getting a significant overhaul that would allow users to create automations using natural language. Instead of manually building workflows step by step, users can describe what they want to happen; Gurman's example has a user setting up a routine that automatically starts a music playlist and sends a spouse an ETA when they begin driving home from work.

Bloomberg previously reporte ond AI-created wallpapers, a systemwide grammar checker for text input, and a revamped Image Playground app offering improved quality for AI-generated pictures and Genmoji custom emoji. See Bloomberg's full report for more information.Related Roundup: iOS 27Tags: Apple Intelligence, Bloomberg, iOS 27, Mark Gurman, Photos
This article, "First Look at AI Camera and Photos Features in iOS 27 Revealed" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Indian CERT urges firms to contain exploited internet-facing flaws within 12 hours

India’s cybersecurity agency, CERT-In, has urged organizations to patch, mitigate, or isolate known exploited vulnerabilities affecting internet-facing “crown jewel” systems within 12 hours where feasible, warning that AI-assisted attacks are dramatically compressing the time between vulnerability disclosure and exploitation.
The recommendation, part of a sweeping new CERT-In blueprint on defending against AI-assisted cyber exploitation, signals a significant escalation in expectations around enterprise vulnerability management, exposure reduction, and operational resilience.
The 38-page framework also recommends one-day remediation for critical externally exposed vulnerabilities, three days for critical internal vulnerabilities affecting high-value systems, and five days for high-severity flaws based on risk prioritization.
CERT-In said threat actors are increasingly using AI to accelerate reconnaissance, vulnerability discovery, phishing, malware generation, and automated exploitation workflows.
“Exploitation timelines are reducing significantly,” the agency warned in the advisory, adding that attacks are expected to become “increasingly autonomous.”
An operationally disruptive target
Security analysts said the headline 12-hour expectation is likely to force enterprises to rethink traditional weekly or monthly patching cycles, but cautioned that the guidance is more nuanced than a blanket patch mandate.
“The 12-hour window is the outlier, realistic only as a containment target on a narrow set of exposed assets, never as a patch-completion target across sprawling estates burdened by fragmented infrastructure, layered approvals, outsourced operations, and legacy dependency,” said Sanchit Vir Gogia, chief analyst at Greyhound Research.
Gogia said the blueprint’s tiered approach is more significant than the headline remediation clock itself because it ties response timelines to exposure and operational criticality rather than applying a uniform patching mandate across all systems.
“The five-day high-severity window is comfortable for most enterprises. The three-day critical-internal window is where the pressure actually bites,” he said, particularly in sectors such as finance, telecom, healthcare, and operational technology environments where uptime concerns complicate rapid change management.
Apeksha Kaushik, senior principal analyst at Gartner, said the biggest challenge for many organizations will not necessarily be deploying patches, but achieving the operational maturity needed for rapid exposure management.
“The primary barriers are not just technical, but operational. Most organizations lack real-time asset visibility, automated vulnerability prioritization, and cross-functional incident response playbooks,” Kaushik said.
“The most acute struggles will be in asset discovery, risk-based prioritization, and orchestrating rapid response across silos,” she added.
From vulnerability management to exposure management
The blueprint repeatedly emphasizes that traditional periodic security assessments are becoming insufficient against AI-enabled attacks capable of rapidly weaponizing newly disclosed flaws.
Instead, CERT-In is pushing organizations toward continuous exposure management, threat-informed defense, continuous monitoring, and adversarial testing.
Notably, the framework leans heavily on temporary mitigations, including isolation, access restrictions, WAF/API protections, enhanced monitoring, and compensating controls when immediate patching is not possible.
Analysts said that approach makes the timelines more achievable operationally, but also shifts the burden onto asset visibility and exposure intelligence.
“Compensating controls do make the timelines more workable. They also remove every excuse,” Gogia said. “If you cannot isolate, restrict, or monitor quickly, the problem was never patch cadence. The problem is that you do not know your own exposure.”
Kaushik similarly said the guidance effectively pushes organizations toward more mature exposure management capabilities.
“Organizations must be able to rapidly identify affected assets, assess risk, and deploy effective interim controls,” she said, adding that enterprises lacking mature asset inventories, segmentation, and monitoring capabilities will struggle to operationalize the guidance at scale.
The blueprint also calls for continuous vulnerability assessments, AI-assisted security testing, adversarial simulations, penetration testing, and red teaming exercises.
A preview of future global standards?
Analysts said CERT-In’s remediation expectations are among the most aggressive currently issued by a national cyber agency and may influence broader international vulnerability-management practices as AI compresses attacker timelines globally.
“CERT-In has done something the West has largely avoided: it has set standing clocks by asset category rather than deadlines by individual vulnerability,” Gogia said.
He contrasted the framework with CISA’s Known Exploited Vulnerabilities (KEV) program, which typically uses vulnerability-specific remediation deadlines rather than persistent enterprise-wide remediation clocks.
“The fixed-clock model looks aggressive today because the rest of the world has not caught up, not because it is reading the threat wrongly,” Gogia said.
Kaushik said the framework could create operational challenges for multinationals whose global service-level agreements are less stringent than India’s expectations. “For providers, this may create a compliance gap where internal SLAs are less stringent than India’s requirements, necessitating a reassessment of global patching and mitigation processes,” she said.
View the full article
CSOonline

iOS 27's New Siri App and 'Search or Ask' Feature Leaked in Screenshots

Bloomberg's Mark Gurman today shared a series of re-created screenshots that reveal iOS 27's all-new Siri app and "Search or Ask" functionality.


With the Siri app, iPhone, iPad, and Mac users will be able to interact with Apple's assistant in both text and voice modes. The app will allow for back-and-forth conversations, complete with messaging history that can be set to expire. Overall, the Siri app would function similarly to chatbots like ChatGPT, Gemini, and Claude.

Siri app:


Search or Ask:

More details to follow.Related Roundup: iOS 27Tags: Bloomberg, Mark Gurman, Siri
This article, "iOS 27's New Siri App and 'Search or Ask' Feature Leaked in Screenshots" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

GlassWorm falls, but the repo problem is far from solved

Taking down a sprawling malware operation once signaled progress in securing the open-source ecosystem. Now, it barely registers. The GlassWorm campaign disruption comes at a moment when attackers can quickly reconstitute, and defenders are increasingly grappling with a new challenge: distinguishing real threats from automated noise.
“I think coordinated actions, like GlassWorm, can sever control, significantly increase attacker costs, buy time for remediation, and signal the possibility of a fightback,” said Agnidipta Sarkar, chief evangelist at ColorTokens. “But most takedowns are temporary actions in a long fight.”
The CrowdStrike-led takedown, conducted alongside Google and the Shadowserver Foundation, disrupted infrastructure linked to the campaign that had poisoned hundreds of repositories with malicious packages targeting developers.
A day after the takedown, in an independent development, the OSV database withdrew 157 malware reports after maintainers determined the submissions were likely automated false positives.
Takedowns help, but analysts question long-term impact
The takedown happened on May 26, at 14:00 UTC, with CrowdStrike confirming the operation to have struck down “all four of GlassWorm’s command-and-control (C2) channels simultaneously”. This reportedly helped sever the botnet operators from their infected machines, blocking them from pushing out new malware.
CrowdStrike described the GlassWorm operation as targeting infrastructure used to distribute malware through developer-focused repositories, an increasingly popular attack vector as adversaries chase CI/CD access, developer credentials, and downstream enterprise environments.
GlassWorm was a cross-platform operation affecting Windows, macOS, and Linux systems, with trojanized VSCode extensions and compromised npm and Python packages for information and credential harvesting.
“As part of our disruption efforts, we are working with partners to bring more pain to attackers, especially when we see them abusing our products or targeting our users,” said Google Threat Intelligence Group’s (GTIG) chief analyst, John Hultquist, in an X post.
Still, the broader economics of repository abuse remain unchanged. Open-source ecosystems continue to offer attackers low-cost distribution, massive reach, and relatively weak identity verification compared to traditional software distribution channels. That means operators behind campaigns like GlassWorm can often reappear quickly under new accounts, domains, or package names.
“It is disruption, not eradication,” Sarkar warned. “To build resilience after a takedown, defenders should prioritize rapid post-takedown scanning to detect the reemergence of malicious artifacts across related repositories and distribution platforms. ”
They should then establish granular micro-perimeters, build capabilities to contain propagation across workloads, endpoints, IT/OT/IoT/cloud assets, and limit the blast radius of supply-chain compromises (e.g., a poisoned npm package or a GitHub workflow stealing creds can’t easily pivot).
Sarkar advised developers and organizations to establish “granular micro-perimeters,” build capabilities to contain propagation across workloads, and limit the blast radius of supply-chain compromises.
AI False positives are becoming part of the supply chain problem
If GlassWorm highlights the persistence of real malware campaigns, the OSV withdrawal incident exposed a parallel issue affecting the open-source software (OSS) supply chain. It is the growing reliability surrounding automated security reporting.
The withdrawal of 157 malware reports believed to be AI-generated false positives matters, especially when it includes packages like FastAPI v0.136.3. FastAPI is a heavily adopted Python framework powering production APIs, AI services, and cloud-native applications across industries. Even a few days of false flagging can trigger costly deployment delays, CI/CD disruptions, and hours of development time in isolating legitimate software.
“I would recommend that enterprises be concerned enough about signal-to-noise problems to consider remedial measures, as automation erodes trust in defensive tools,” Sarkar said. “Unless you have a highly microsegmented enterprise, noise wastes analyst time, slows velocity, and risks missing sophisticated attacks amid fatigue.”
In 2026, with AI-assisted malware and reporting both accelerating and rising false positives in SAST/SCA tools, defensive automation is getting asymmetrically compounded by supply-chain volume, he noted.
In a blog post, Socket called bad OSV records particularly dangerous as the popular database gets rapidly carried through dependency scanners, CI checks, registry controls, SBOM tools, dashboards, and internal policy systems.
All hope is not lost, though, as newer tools promise lower reliance on AI for hunting dependency vulnerabilities. CVE Lite CLI, a light-weight, JavaScript and TypeScript dependency vulnerability scanner, is offering developers a way to know dependency risks while they are still writing code, much earlier than failing automated scanners in CI pipelines.
View the full article
CSOonline

How to Evolve from a Human-Led SOC to an Agentic SOC Model

Moving toward an agentic SOC is not a single platform decision. It is an operating model shift that changes how triage, investigation, and response move through the SOC. In this blog, we look at how organizations can evolve from a human-led SOC to an agentic SOC model in a phased, controlled way that improves efficiency without losing oversight.
View the full article
compuquip

The AI governance imperative you can’t afford to ignore

CIOs rushing to roll out AI agents without real visibility into their decision-making processes are flirting with disaster.
According to AI experts, deploying agents without observability processes and tools creates a ticking time bomb with the potential for huge negative consequences.
Many companies are deploying AI agents and expecting them to increase productivity with little human intervention, observes T.J. Marlin, CEO of AI security firm Guardrail Technologies. That’s the wrong approach, he says. Instead, IT teams need to keep a close eye on agents and adjust policies and practices throughout the agentic process.
“It’s not just set it and forget it like a crock pot,” he says. “You don’t put it in the kitchen in the morning with the chicken inside and come back at night and have a great dinner. The organizations doing that are going to be on the front page because they just had some terrible thing happen to them.”
Many organizations are rapidly deploying agents because of a fear of missing out, while not understanding the nuances of the technology, Marlin says. Some IT leaders seem to compare agents to robotic process automation, when RPA results are far more deterministic, he adds.
“There’s a talent shortage and a knowledge shortage and people are building at pace without checking whether it’s correct and it’s operating as expected,” he says. “Those are all the hallmarks of the worst disasters that I’ve seen across my career.”
A recent report from agent governance vendor TrueFoundry puts numbers behind fears of unregulated agents. A survey of more than 200 enterprise AI leaders found that 54% of organizations represented can’t fully trace what their agents are doing and 56% have no centralized agent control or governance layer.
While TrueFoundry has an interest in driving agent governance forward, many other AI experts see the same problems.
Governing blind
Difficulties with governance and observability are major impediments to the deployment of productive agents, and many organizations are deploying agents without creating a centralized list of them, says Mahesh Kumar Goyal, senior data and AI expert at Google.
“Most enterprises have no inventory of the agents already running in production — they’re trying to govern what they can’t see,” he says.
In addition, traditional SIEM and EDR security tools were built to spot human anomalies, not rogue agents, he notes. “An agent running code perfectly 10,000 times in a row looks normal even if it’s been hijacked,” he says.
Running fully autonomous agents is not a good idea, he adds, and organizations need to think about least-privilege scoped tool permissions, policy enforcement layers that mediate every prompt and tool call, and end-to-end tracing that stitches prompts, tool calls, and downstream actions into one auditable trail.
“The financial system doesn’t run on trust; it runs on auditability, reconciliation, and circuit breakers,” Goyal says. “Agents will mature the same way. Tiered autonomy is the realistic answer: free rein on low-stakes tasks, human-in-the-loop on consequential ones.”
Part of the problem is that agents have upended the models used to determine whether traditional software was running correctly, adds Adel El Hallak, vice president of AI software at Nvidia. With traditional software, QA and security professionals could look at the code to debug problems, but agents make decisions in the runtime environment of an AI model.
The source of truth for agents resides in the traces, the records of the execution flow, not in the code, he adds. Collecting traces ­­— in essence, detailed logs — is a start toward agent governance, but organizations need to be able to act on the information, he says.
“For you to trust something, it has to be transparent, and observability is foundational to transparency,” El Hallak adds. “But just observing is not enough. We need to be able to take those signals and turn them into something actionable.”
Agent governance goes beyond observability to allow organizations to test and fine-tune agents continuously, he says. The tools are out there, with companies like Nvidia building their own internal governance frameworks, and several other vendors offering agent observability and governance tools, he notes.
“It’s not enough to just have the behavioral data, to capture the feedback data,” he says. “The system should allow me to annotate, change, augment, or create additional feedback data, and then I can use that data to improve my agent as a whole.”
The governance bottleneck
At the same time, many companies moving into agent governance have found it can be a huge bottleneck if done wrong, says Nirmal Ganesh, senior director of product management for agentic workflow automation at cloud storage vendor Box.
“I don’t believe we are past the hard part yet in terms of deploying agents in the enterprise,” he says. “Most companies are not yet good at those, and far fewer of them have gotten good at running them at scale with agent governance and observability.”
Ganesh sees several problems, including agents running without clear permission models. “If an agent can see more than a person or access more than a person’s permission on content or data, that’s an incident is waiting to happen,” he says.
However, some early agent governance models don’t scale. Some IT teams have defaulted to a position of humans needing to approve every agent output because that’s the safest option, he says.
“In reality, this is rebuilding manual process with more checkpoints or suggestion points,” Ganesh says. “At a high volume, governance is your bottleneck to scale and no longer your safely net.”
Organizations need observability and governance processes in place that are both scalable and comprehensive, he adds. Agent ROI will come from strong guardrails, clear permission models, and clear human-in-the-loop involvement, he says.
“Every mature automation needs ongoing observability — workflows change, policies change, decisions change, new use cases show up,” he says. “Human intervention is always needed for what changes over time, but we need less intervention for known paths and more focus on exception handling and governance fine-tuning.”
Observing output is not enough
Governance can’t just focus on agent output, adds Marcelo Lorenzetti, founder and CAIO at legal services AI vendor SavvyLex.
“The biggest challenge is not simply whether an agent produces a good answer,” he says. “It is whether the organization can prove what the agent accessed, what instructions it followed, what tools it invoked, what decisions it made, where a human intervened, and whether it stayed within authorized boundaries.”
Without a full level of runtime visibility, companies are left with screenshots, logs, and after-the-fact explanations that may not meet legal, compliance, or security requirements, he says.
Agents should be continuously verified instead of fully trusted, he adds, with governance engineered into the agent architecture itself. Governance should include role-based access, policy-bound execution, human approval thresholds, source and tool provenance, immutable activity records, confidence scoring, exception handling, and clear escalation paths when an agent reaches the edge of its authority, he recommends.
“Observability should not be limited to whether the model responded,” Lorenzetti says. “It should show the full decision path from input to action.”
AI agents have shifted the governance model that’s needed, he adds.
“The core problem is that many companies are moving from AI that answers questions to AI that takes actions, but their governance models are still built for passive tools, not autonomous workflows,” he says.
View the full article
CSOonline

What the industrialization of exploitation means for defenders

For decades, cybersecurity was a battle of skill. Elite attackers versus elite defenders. The rules of engagement were understood, even if the playing field wasn’t level. If you hired better analysts and bought better tools, hopefully you hardened your systems well enough and built detection capabilities that wore out the adversary’s patience.
That era is over, and most security programs haven’t fully processed what replaced it. Adversarial AI has industrialized exploitation. What once required a coordinated team of technically sophisticated threat actors to manage reconnaissance, weaponization, lateral movement and persistence can now be executed autonomously, at machine speed, against thousands of environments simultaneously. Threat actors no longer need deep technical expertise. They need compute, capital and access to AI tooling — all of which are commoditized.
Think about what your team used to rely on. Attackers left clues that telegraphed their presence – patterns you could learn, signatures you could catch and their campaigns moved slowly enough to track. That’s gone. Reconnaissance that took days now takes minutes. The attacks your tools were trained to recognize are being rewritten on the fly. And the coordinated human teams that once limited how many targets an adversary could hit at once? They can now be easily outmaneuvered by a single actor with the right AI tooling. Your architecture was designed for a threat that no longer exists.
The problem is structural
The gaps AI-enabled adversaries are exploiting aren’t primarily operational failures. They’re architectural ones. As enterprise environments expanded across cloud, OT, identity infrastructure and third-party integrations, security organizations responded by layering tools. Each new surface area got a new control, a new scanner, a new dashboard. This has created a security architecture that’s simultaneously complex and fragmented — generating enormous volumes of signal while producing limited clarity about where the actual risk lives.
The specific failure modes are familiar to anyone who has worked through a real breach investigation. Controls that don’t share context mean a vulnerability scanner can flag a misconfiguration, an identity tool can flag an overprivileged account and an endpoint platform can generate an alert — none of them are able to answer the question an attacker has already answered: Can these exposures be chained into a viable path to something critical?
Visibility across hybrid and multi-cloud environments remains patchwork at best; attackers move freely across boundaries that defenders frequently can’t see across. Identity exposure — overprivileged service accounts, stale credentials, misconfigured trust relationships — creates lateral movement pathways that go undetected until someone is already deep inside the environment. Alert overload causes security teams to spend disproportionate time on findings with no realistic exploitation path.
None of this surprises working security professionals. What’s less widely acknowledged is that it’s not a resourcing problem. More analysts and more siloed tools, layered onto a fragmented architecture, produce more of the same. Security tools are built to detect and flag. They weren’t built to show you what an attacker sees when looking at your environment.
Attackers have already leveraged automation to extend their reach. AI will enable them to exploit attack paths with unprecedented speed. So, as clichéd as it sounds, defenders need to put themselves in the shoes of attackers and adjust their approach from there.
How defenders can change the equation
That mindset shift starts with asking different questions. Most security programs are built around “what vulnerabilities exist?” The better question is “what can an attacker actually do with what’s in my environment right now?”
That reframing has real consequences for how programs are run. Incident response speed matters, but it’s a downstream variable. The upstream question is how to make incidents caused by structural gaps and flaws less likely — which requires understanding your environment the way an attacker would, as a network of relationships that can be chained, not as a collection of independent assets and controls. Most security teams have never mapped their environment from that vantage point. Most attackers have.
It also means prioritizing remediation by real exploitability rather than CVSS score or asset criticality in isolation. This is Exposure Management 101 — the “EM” in Gartner’s Continuous Threat Exposure Management framework, which provides a structure for replacing broken vulnerability management processes. Exposure Management operationalizes the “think like an attacker” ethos at scale.
Security programs that prioritize real exploitability are working on the right problem. The 2025 Verizon DBIR found that the median time for edge device vulnerabilities to be mass-exploited was zero days, while organizations took a median of 32 days to fully remediate them. And separately, the average time to patch across 17 high-profile edge device CVEs was 209 days. You can’t close that gap by triaging everything equally.
The defender’s actual advantage: Know thy environment
There’s a version of the current threat landscape that leads to fatalism.  Why invest in a fight you’re structurally losing? It’s easy to go there, but it’s the wrong read. Ultimately, I believe that defense will become equally automated — a true battle of the machines.
But even before we get there, defenders have a structural advantage that no amount of adversarial AI eliminates: They operate inside the environment they’re protecting. They can see the full topology, the identity relationships, the compensating controls, the critical assets. An attacker, however sophisticated the tooling, has to discover all of that from the outside. Defenders already know it. At least they should.
Most organizations have the underlying data to understand their own exposures. The challenge is synthesizing it into something actionable — seeing on a continuous basis what an attacker would see, and which paths actually lead somewhere dangerous.
Start with visibility that actually crosses the boundaries your tool stack has carved out over years of reactive purchasing. Get serious about prioritization based on what’s genuinely exploitable in your environment, not what scores highest on a spreadsheet. And stop conflating compliance-driven tests with your current risk posture — they tell you what things looked like last quarter, not today.
The conversations CISOs should be having at the board level should focus on whether the program running today can flag when an AI-empowered attacker has a clear path to the company’s crown jewels.
The industrialization of exploitation is a genuine shift in the adversary’s economics and logistics. But the structure of the problem hasn’t changed. Defenders who understand their own environment better than attackers — and who build their programs around that advantage — are in a stronger position than the threat headlines suggest.
Are you leveraging the defender’s advantage?
The fast way to know this is to have your team answer the following questions:
How many critical corporate assets have a validated attack path from an internet-facing entry point? How has that number changed quarter-over-quarter? What percentage of our remediation effort closed an actual path versus a theoretical finding? Do we know the ways an attacker could create an attack path to our critical assets? Are we continuously assessing all of the possible attack paths to our critical assets?” Then, if you don’t like the answers, it’s time to revisit your control architecture. The best way to avoid cyber disruption from adversarial AI is to fix the structural problems so those attack paths aren’t realized in the first place.
Carpe Diem!
 
This article is published as part of the Foundry Expert Contributor Network.
Want to join?
View the full article
CSOonline

Mastering DevOps Culture: A Guide for Developers and System Administrators

Introduction
In the current landscape of software engineering, technical proficiency is the baseline expectation. If you know how to write code, configure servers, or manage cloud infrastructure, you have entered the game. However, many engineers find that their career trajectory plateaus despite having advanced technical knowledge. They master the tools but struggle to lead, innovate, or advance into senior roles. This is where the missing piece—DevOps culture—becomes critical.
Many professionals mistakenly view DevOps as a collection of tools like Jenkins, Kubernetes, or Docker. In reality, DevOps is a philosophy. It is a way of working that emphasizes communication, shared responsibility, and continuous improvement. Without understanding this culture, you are just performing tasks; with it, you are building systems and, more importantly, a lasting career.
The demand for professionals who understand this shift is rising rapidly. Companies are no longer looking for siloed experts. They need engineers who can bridge the gap between development and operations. By cultivating this mindset, you position yourself as a force multiplier within your organization.
For those looking to deepen their understanding of these practices, platforms like DevOpsSchool provide essential resources, training, and a community focus that helps bridge the gap between theoretical knowledge and real-world engineering culture. Developing this mindset is not just about company goals; it is about your own professional longevity.
What Is DevOps Culture?
At its core, DevOps culture is the human side of software engineering. It is the practice of breaking down the traditional walls between the people who write code and the people who maintain it.
Think of it as a shift from “my job ends here” to “we succeed together.” It is a mindset that prioritizes empathy, transparency, and a relentless focus on the end-user. If a developer pushes code that breaks the production environment, a DevOps culture does not look for a scapegoat. Instead, it looks for the gap in the process that allowed the failure, fixes it, and ensures it does not happen again.
The Key Ingredients
Collaboration: Actively seeking out different perspectives to solve problems. Shared Responsibility: Accepting that the health of the application is a team effort, not an individual burden. Continuous Improvement: Always looking for ways to make the process smoother, faster, and more reliable. Why DevOps Culture Matters in Modern Engineering
Modern software engineering is fast-paced. Requirements change, cloud services evolve, and users demand instant reliability. If your team operates in silos—where developers throw code over a wall to an operations team—you will inevitably face bottlenecks.
DevOps culture matters because it creates a safety net for innovation. When teams collaborate, they identify security flaws earlier, catch bugs faster, and reduce the time it takes to get features into the hands of users. This is not just about efficiency; it is about building software that actually works and improves the lives of those using it.
Example in the Workplace
Consider a scenario where an application crashes during a high-traffic sale. In a traditional environment, the developers might blame the server config, while the operations team blames the code. In a DevOps culture, the team acts as one unit. They immediately look at the monitoring dashboards, correlate the crash with recent deployments, and work together to roll back or patch the issue within minutes. This speed is only possible when the culture is built on trust and shared data.
Core Principles of DevOps Culture
PrincipleCareer BenefitCollaborationBuilds strong cross-functional relationships and networking skills.AutomationFrees up your time to focus on complex, high-value tasks.Continuous FeedbackSharpens your ability to learn from mistakes and improve.Shared OwnershipEstablishes your reputation as a reliable and accountable leader.Learning MindsetKeeps your skills relevant in a rapidly changing market. Collaboration
Collaboration is the bedrock of this culture. It is not just about meetings; it is about open communication channels. An engineer who talks to the QA team early in the design phase is far more valuable than one who hides behind a monitor.
Automation
Automation is the mechanism that allows for scale. If you are doing a task more than once, it should be automated. This requires you to move beyond manual intervention and toward building systems that manage themselves.
Continuous Feedback
This involves using metrics and monitoring to understand how your software performs in the real world. You are not guessing anymore; you are using data to drive your next move.
How DevOps Culture Impacts Your Career
If you want to move from an individual contributor to a lead, architect, or manager, you must embrace this culture. The impact on your career is tangible.
Faster Career Growth
Professionals who understand DevOps culture are viewed as problem solvers rather than ticket-closers. You start being involved in architectural discussions earlier because stakeholders trust your collaborative approach.
Cross-Functional Expertise
By working across teams—from security and development to operations—you gain a bird’s-eye view of the product lifecycle. This knowledge is invaluable for roles like SRE or Solutions Architect.
Leadership Opportunities
Leadership is not just about managing people; it is about influencing processes. By championing automation and collaboration, you naturally step into positions of influence, making you the person management turns to when they need to scale a team.
Collaboration Skills in DevOps Careers
Collaboration is often mistaken for just being “friendly.” In the context of engineering, it is a technical skill. It requires the ability to explain complex problems to non-technical stakeholders and the humility to accept feedback on your code.
Engineering Scenario
Imagine you are building a new deployment pipeline. Instead of building it in isolation, you reach out to the security team to understand their compliance requirements first. By doing this, you avoid having your pipeline rejected later. That is collaboration as a competitive advantage. It saves time and builds professional political capital.
Automation Mindset and Career Growth
The automation mindset is about looking at every manual task as a technical debt. If you are manually patching servers or copy-pasting configurations, you are not using your brain to its full potential.
Practical Productivity
When you automate a process, you are effectively creating a product. You are taking your expertise and embedding it into the system. This creates a legacy of your work that continues to provide value long after you have moved on to new projects. This is what separates an average engineer from a top-tier performer.
Ownership and Accountability in DevOps Culture
The mantra “you build it, you run it” is the ultimate test of ownership. When you are responsible for the code you deploy, you naturally write better, more maintainable, and more robust code.
Workplace Example
Think of the engineer who deploys a database migration and goes home immediately. Compare that to the engineer who stays online to ensure the metrics remain stable after the deployment. The latter displays a level of accountability that is immediately noticed by leadership. This accountability is the fastest way to earn trust.
Continuous Learning and DevOps Success
The landscape of technology changes every six months. If you rely on the knowledge you gained in college or your first job, you will become obsolete. DevOps culture fosters a “beginner’s mind,” where learning is not a chore but a daily habit.
Whether it is learning to secure a container, understanding cloud cost optimization, or mastering infrastructure-as-code, your value is tied to your ability to learn. This is why continuous improvement is not just a company goal—it is your personal insurance policy.
DevOps Culture vs Traditional IT Culture
AreaTraditional ITDevOps CultureCollaborationSiloed, team-specific knowledge.Open, shared knowledge across teams.Delivery SpeedSlow, prone to release friction.Rapid, consistent, and automated.AccountabilityBlame-driven (“not my fault”).Ownership-driven (“how do we fix this?”).LearningStagnant, based on legacy processes.Constant, based on feedback loops.FeedbackReactive (post-mortem only).Proactive (real-time monitoring). Real-World Example: Engineer Without DevOps Culture
Consider Alex, a developer who is brilliant at writing code. Alex is fast, writes clean syntax, and never misses a deadline. However, Alex treats the operations team as an annoyance, views documentation as a waste of time, and ignores the security implications of his dependencies.
When a bug hits production, Alex is nowhere to be found, claiming, “I wrote the code; the ops team manages the server.” Over time, Alex’s code becomes brittle, team members avoid working with him, and his career stagnates because he has not learned how to scale himself or his work.
Real-World Example: Engineer With DevOps Culture
Now consider Sam. Sam writes good code but always documents the “why” behind his logic. When he builds a feature, he collaborates with the QA and SRE teams to ensure the release plan is solid. When a production issue happens, Sam is the first to open the logs and help the team diagnose the root cause, not to place blame, but to solve the issue.
Sam is known as a reliable partner. Because of this culture, Sam is offered leadership opportunities, asked to mentor juniors, and is often the first person considered for promotions. Sam understands that his success is tied to the success of the whole team.
Soft Skills That Matter in DevOps Culture
Communication: Can you articulate the value of a technical change to a project manager? Teamwork: Do you elevate the performance of those around you? Problem-solving: Do you seek root causes or just symptom relief? Adaptability: Are you comfortable changing course when data suggests it? Ownership: Do you treat the company’s goals as your own? Common Beginner Misunderstandings
“DevOps is only tools.” False. Tools are the vehicle, not the destination. “Culture does not matter if I code well.” False. Technical skills get you hired; soft skills get you promoted. “Automation will replace me.” False. Automation allows you to work on higher-level, more rewarding problems. “Only DevOps Engineers need this.” False. Every developer and system admin needs this mindset to remain competitive. Best Practices to Build a DevOps Mindset
Ask Questions: Do not just execute tasks. Ask why the process exists. Practice Automation: Start by automating your smallest, most repetitive daily task. Build Relationships: Shadow someone from a different department (e.g., Security or QA) for a day. Share Knowledge: Write documentation for the things you learn. Stay Curious: Follow trends, read engineering blogs, and keep your toolset updated. Role of Leadership in DevOps Culture
Leadership sets the tone. If managers value “firefighting” over “preventative engineering,” the team will never adopt a DevOps culture. Leaders must empower their teams by allowing them to fail fast in safe environments, providing the time for learning, and rewarding collaboration over individual heroics.
Role of DevOpsSchool in Building DevOps Skills and Culture
Building a career in this field requires more than reading articles; it requires structured guidance and practical application. DevOpsSchool supports this journey by offering environments where you can practice CI/CD, understand cloud infrastructure, and learn the collaborative practices that define this culture. By engaging with these resources, you transition from someone who just knows the theory to someone who has actually applied the engineering mindset in a real-world context.
Career Opportunities Influenced by DevOps Culture
DevOps Engineer: Bridging development and operations. Cloud Engineer: Managing scalable infrastructure. SRE (Site Reliability Engineer): Focusing on uptime and reliability. Platform Engineer: Building internal tools for other developers. Automation Engineer: Specializing in process optimization. Engineering Leader: Managing teams and culture. Industries Benefiting from DevOps Culture
Banking & Finance: Where security and reliability are non-negotiable. Healthcare: Where data integrity and uptime save lives. SaaS Platforms: Where rapid feature delivery is a competitive advantage. E-Commerce: Where performance directly correlates to revenue. Telecom: Where massive scale and automation are essential. Future of DevOps Culture
The future is shifting toward “Platform Engineering” and AI-assisted operations. The fundamental principle—collaboration and shared responsibility—remains unchanged. As AI begins to handle more of the manual tasks, the value of the human engineer will shift toward system architecture, strategic thinking, and fostering the cultural environment that allows technology to thrive.
FAQs
What is DevOps culture exactly?It is a collaborative work style that removes silos between teams, focusing on shared responsibility and continuous improvement. Why does DevOps culture matter for my career?It accelerates your growth, makes you a more effective team player, and prepares you for leadership roles. Can beginners learn DevOps culture?Yes. It is a mindset that can be cultivated by anyone willing to prioritize collaboration and learning. Is DevOps only about automation tools?No, tools enable the culture, but the culture is about how people interact and work together. What skills matter most in this culture?Communication, adaptability, ownership, and a deep understanding of the entire product lifecycle. Does DevOps culture help career growth?Absolutely, as it builds the soft skills and architectural thinking required for senior and leadership positions. Is collaboration important in DevOps?It is the foundation. Without it, you are just working in a different, more technical silo. How can I develop a DevOps mindset?Start by automating small tasks, talking to other teams, and focusing on the “why” behind your work. Do developers need DevOps culture?Yes. It helps developers write code that is easier to deploy, monitor, and maintain. Does DevOps culture increase productivity?Yes, by removing friction, reducing manual errors, and speeding up feedback loops. Is it only for large enterprises?No, teams of any size benefit from shared ownership and automated processes. How do I handle team friction in a DevOps transition?Focus on empathy, clearly communicating the benefits, and leading by example. Is DevOps culture about SRE?SRE is a specific way to implement DevOps principles, but the culture itself is broader. What is the biggest challenge?Changing human behavior and overcoming the “this is how we have always done it” mentality. Where should I start my learning journey?Focus on mastering one domain, like cloud or CI/CD, while actively seeking collaboration opportunities at work. Final Thoughts
Technical skills are your currency, but your cultural mindset is your investment strategy. You can spend your career focusing solely on the “next big framework,” or you can develop the habits, collaboration skills, and engineering maturity that define a true DevOps professional.
When you start taking ownership, automating the mundane, and looking for ways to support your teammates, you stop being just another engineer. You become the engineer that teams rely on, that leadership values, and that consistently moves forward. Culture is not just a company buzzword; it is the most reliable path to a successful and fulfilling career in modern technology. Start building that mindset today.
View the full article
reporter

Employees are unknowingly inviting tech support impersonators into firms, says FBI

Online or telephone IT support scams have been tricking employees into downloading or clicking on malware for years. But according to the FBI, one group that targets US-based law firms has recently found success in person, by convincing firms to allow a supposed IT support person into the building, where they insert a storage device into a victim’s computer and install malware or steal data.
This revelation comes from an FBI Flash report this week describing the activities of a gang it calls The Silent Ransom Group (SRG). Other researchers call it Luna Moth, Chatty Spider and UNC3753.
Cybersecurity experts, though, aren’t surprised that employees can be fooled into allowing a stranger to touch their computers.
“The adversary visiting a location in person with a USB key hacking device of some sort has been used for decades, particularly in the banking industry,” said Roger Grimes, CISO advisor at KnowBe4. 
“Usually, it isn’t just a direct download of data, but using the USB storage drive to either monitor password typing, to install remote access software that the hacker can use to come back into the environment remotely, or to install some other sort of hacker malware. It’s so common in the banking industry that they have often added and allowed that scenario — a physical attacker — in their regular penetration testing audits, more so than any other industry.”
Lance Spitzner, director of workforce cybersecurity training at the SANS Institute, said the tactic of getting into a company to use infected USB drives isn’t new, but in his opinion is relatively rare. It’s more common for a threat actor to mail a drive to an employee.
“Having someone physically expose themselves by going into an organization is a risk most cyber attackers are not willing to take,” he said. “The details in the FBI report are pretty limited; I’m guessing if this did happen, an attacker paid someone off to do it for them, perhaps an insider or contractor the company trusted.”
The FBI says SRG actors have been running data theft and extortion operations since at least 2022. Despite its name, the gang doesn’t use ransomware encryption, but typically seeks rapid access to victim systems to steal data. Then they use extortion, through threats of public disclosure or sale of stolen data, to try to get payments.
Historically, the gang gained access to the victim’s network by sending phishing emails purportedly charging small ‘subscription fees;’ to cancel the fake subscription, the victim was instructed to call the threat actor, who then emailed the victim a link that would download remote access software.
New tactic
But since the spring of this year, SRG actors have added a new tactic: Posing as an employee from the victim’s IT department. They either directly call or send phishing emails to urge employees to contact an SRG actor pretending to be their firm’s IT support. While on the phone, the SRG actor asks the employee to grant access to a remote desktop session.
If that fails, SRG sends a threat actor to the victim’s location to physically access their computer and insert a storage device. The excuse: the so-called IT support person needs to image the device, or to create a backup file to address potential impacts from the phishing email. Once the threat actor obtains access to the victim’s device, they minimally escalate privileges and quickly pivot to data exfiltration without encryption.
Their tools include WinSCP (Windows Secure Copy) or a hidden or renamed version of “Rclone” to exfiltrate data. They may also exfiltrate data to internal file sharing platforms such as Google Drive or Microsoft OneDrive. And once it has the firm’s data, the gang will call employees or clients of a victim company to pressure the victim to begin negotiations.
The FBI warns infosec pros that indicators of an SRG attack may include new, unauthorized downloads of system management or remote access tools, including Zoho Assist, Quick Assist, AnyDesk, RustDesk, Syncro, Splashtop, or Atera; unauthorized installation of external hard drives or USB drives on company computers; exfiltration of data to Microsoft OneDrive, Google Drive, or external servers; WinSCP or Rclone connection made to an external IP address; and alerts that data was exfiltrated from the company environment.
The primary things employees need to be trained to watch for are visits from unidentified or unauthorized individuals claiming to be IT support and attempting to access computers, and unsolicited phone calls from individuals falsely claiming to work in their IT department.
The FBI didn’t respond by press time to a request for information on the number of times the gang had tricked an employee into allowing a personal visit. But the tactic by SRG is new enough that the bureau is asking for a copy of the extortion note, the phone number or email account used by the group, transcripts of communications with the threat actor, and any surveillance videos or photos of individuals posing as IT support.
The challenge of security awareness training
Since the beginning of the desktop computer age, CSOs, CIOs and IT department leaders have struggled to find effective security awareness training to fight phishing, IT tech scams, and other social engineering attacks. Law enforcement agencies have had some success in taking gangs offline, but they pop up again.
The threat actors may also be assisted by the fact that employees often don’t know who their IT support staff are, especially if the firm uses a third party external support company.
[Related content: A backgrounder on security awareness training]
Christopher Kayser, head of the Canadian firm Cybercrime Analytics and author of the book Cybercrime Through Social Engineering, said in an interview that often an employee’s first assumption is that an email, text, or voicemail about a serious issue from someone claiming to be from IT is legitimate.
After that, threat actors play on an employee’s willingness to act on a supposedly urgent matter, their obedience to management, or their wish to be helpful. “We have a tendency to trust,” he said.
It doesn’t help that threat actors are willing to share successful tactics with other groups, he added. Nor, he said, does it help that in some organizations, security awareness training doesn’t extend to the top (CEOs) or the bottom (receptionists).
Trust no one
To combat IT support scams, employees need to be trained that any email, text, or voicemail purporting to come from IT that asks for action needs to be verified with an IT manager through an approved process, not by replying to the message or calling a phone number given in the suspect communication, Kayser said. Employees also need to be trained to slow down, and not to respond or act quickly on emails, texts, or voicemails that ask for passwords, multifactor authentication codes, or personal information.
Spitzner added, “security awareness training is one approach to patching vulnerabilities in humans. You need to teach them about the risks of infected or untrusted USB drives and what drives are authorized.  In this case, the problem may have been an untrusted individual gaining access to the victim’s facilities.”
Nick Tausek, lead security automation architect at Swimlane, said the Silent Ransom Group’s attack strategy of leaning into trust says a lot about where extortion is heading. “That makes this especially dangerous for law firms,” he said. “Those environments hold sensitive client records, privileged communications, financial details, and case information. If that data is stolen, the damage does not stop at the victim organization. Clients can be pressured, legal strategies can be exposed, and employees can become targets for follow-up scams.”
The hardest part is that much of this activity can look normal at first glance, he said. Because legitimate tools used by threat actors don’t always trigger alarms, security teams need faster ways to connect unusual behavior across users, devices, cloud storage, and remote access sessions. “When attackers are moving this quickly, delayed detection gives them the advantage,” he said.
Grimes added that defenses should include strong and frequent employee education about physical attacks, disabling USB ports on publicly accessible computers, and other mitigations that prevent the connection of physical storage devices. Microsoft Windows, he pointed out, has had mitigations to prevent the insertion of unauthorized storage devices, including USB sticks, for well over a decade.
In addition, the FBI urges physical and IT security leaders to verify the credentials of anyone accessing company spaces, obtaining copies of each visitor’s ID card, as well as limiting access to sensitive data from less secure networks, such as home computers or the public internet, and developing and communicating policies regarding when and how IT support will communicate and authenticate themselves to employees.
View the full article
CSOonline

Amazon to Acquire Apple's Globalstar Stake in Satellite Deal

Amazon is acquiring Apple satellite provider Globalstar, and it has worked out a deal to take over Apple's stake in the satellite company.


According to an FCC filing noticed by PCMag, Amazon is initiating a merger structure. It will create a new subsidiary called Grapefruit Acquisition Sub II to absorb Globalstar's operations, and then the new subsidiary will acquire Apple's equity.

Apple has 20 percent equity and 20 percent voting interests in Globalstar Licensee LLC, a Globalstar subsidiary that operates the satellite system that delivers satellite connectivity to iPhone users in the U.S. Apple invested $450 million in Globalstar back in 2022 to fund Globalstar's satellite buildout for the Emergency SOS via satellite feature. Apple received 20 percent equity and 85 percent of Globalstar's network capacity in return.

Apple's Globalstar stake required Apple and Amazon to enter into separate negotiations, but the two companies worked out a deal. Amazon's Leo satellite network will power existing iPhone and Apple Watch satellite features like Emergency SOS, Messages via satellite, Find My, and Roadside Assistance via satellite. Apple stands to make more money than it invested, and has secured a partnership with a better-established satellite operator.

Amazon will support iPhone and Apple Watch models that use Globalstar's low Earth orbit constellation, and the two companies will work together on future satellite services running on Amazon's expanded Leo network.

Amazon's Globalstar purchase is said to be worth around $11.57 billion, and it is expected to close in 2027 as long as it earns regulatory approval.Tags: Amazon, Globalstar
This article, "Amazon to Acquire Apple's Globalstar Stake in Satellite Deal" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

YouTube Will Now Automatically Label AI Videos Even When Creators Don't

YouTube today said it is introducing automatic AI detection, with automatic AI labeling applied to videos with "significant photorealistic AI use."


Creators are still expected to manually disclose when realistic AI is used for videos even with the new automatic labeling system. Creators who think their content was incorrectly identified as AI-generated can update the disclosure status in YouTube Studio, but disclosures are permanent for content created with YouTube's AI tools like Veo and Dream Screen or content with C2PA metadata indicating fully generative AI.

YouTube is also improving labeling for AI-generated content, making it clearer when a video has "photorealistic and meaningfully AI altered or generated content."

An AI label will be shown just below the video player and above the description for long-form videos with AI content, and for shorts, the label will appear as an overlay on the video. The updated labeling applies to content that is photorealistic and may fool people, rather than unrealistic content.

Disclosures for content that is "unrealistic, animated, or slightly altered" will continue to be tucked in the expanded description of the video.

Separately, YouTube added a new customizable content feed that users can shape based on interests, moods, or preferred topics. Users can type in a custom prompt covering what they want to see, and a content feed will be generated.

Custom content feeds have been in testing since November, but they are now rolling out to signed-in viewers in the U.S. on the YouTube mobile app and desktop. YouTube search and watch history must be turned on for the feature to work.Tag: YouTube
This article, "YouTube Will Now Automatically Label AI Videos Even When Creators Don't" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Another IT governance headache: AI-enabled sanction evasion

Over the next three to five years, both governments and the private sector will need to rapidly adapt identification and mitigation protocols as adversaries move from AI-assisted to AI-enabled sanctions evasion and proliferation financing (PF), a new research paper warns.
The report, Algorithms of Evasion: The Rise of AI-Enabled Proliferation Financing, from the Royal United Services Institute (RUSI), a UK-based defense and security think tank, defines PF as the use of funds or financial services to acquire, develop or otherwise deal in weapons of mass destruction (WMD). It states, “North Korea and Iran are now developing and deploying AI models to aid with sanctions evasion activities.”
Key findings include the fact that AI is now capable of mass producing high-quality fraudulent documents, as well as automating what the report describes as “the administrative minutia of managing extensive shell company  networks.” AI powered systems, it states, can also “analyze blockchain patterns in real time to dynamically adjust cryptocurrency mixing strategies, effectively evading detection tools.”
In addition, it says, “[tools such as generative AI] which can produce sophisticated fraudulent identification documents, for example, have helped North Korea perpetrate phishing attacks against Western companies.”
Dr. Aaron Arnold, senior associate fellow with the Centre for Finance and Security at RUSI, who authored the paper, said in an email that what prompted it was an uptick over the last year in North Korea’s use of AI to facilitate and enhance its cyber operations, in the form of phishing schemes designed to generate revenue for the country’s ballistic missile and nuclear weapons programs.
He advised enterprise IT managers who need to protect their organizations from becoming victims of sanction evasion activities that “[it] means largely adapting to a landscape where traditional human-focused security boundaries are being bypassed by automated technologies.”
For IT managers, said Arnold, “this might entail incorporating defensive AI, the use of behavior-based analytics, using ‘circuit breakers’ when there is heavy use of API or MCPs, updating personnel training, and hardening identity verification, especially for any remote hiring.” 
Distinction between AI-assisted and AI-enabled activity is ‘central’
Sanchit Vir Gogia, chief analyst at Greyhound Research, said that the RUSI report matters “because it names the right structural shift. AI is not creating sanctions evasion from thin air, it is compressing and scaling methods that already work.”
He pointed out that none of the sanction-evading techniques such as fraudulent documents, synthetic identities, shell companies, hidden beneficial ownership, crypto laundering, and others are new. “What changes is the speed, quality, volume and coordination with which these methods can now be assembled,” he said.
According to Gogia, “the distinction between AI-assisted and AI-enabled activity is central. AI-assisted evasion uses AI for discrete tasks: writing a better email, producing a cleaner document, generating a stronger false profile, translating a pitch, summarizing regulations or preparing a plausible job application. AI-enabled evasion is more serious.”
A ‘structural asymmetry’
This tactic, he said, “begins to coordinate the system itself. It links identity, documents, ownership structures, payment routes, cloud access, crypto wallets, API calls and timing. The difference is not whether AI helps someone fake a document. The difference is whether AI begins to orchestrate the deception.”
That is why the report’s findings should worry enterprise leaders, he noted: “Many organizations still assume the bad actor is mostly human, mostly linear and mostly slow. That assumption is expiring. AI lets adversaries run more attempts, with fewer errors, across more channels, in more languages, with better paperwork and greater patience than most enterprise review processes can absorb. This is not a tale of genius criminals discovering magic. It is the story of ordinary controls meeting industrialized plausibility.”
The evidence today, he pointed out, is strongest around tactics such as identity fraud, document fraud, synthetic personas, remote-worker deception, phishing, social engineering, crypto obfuscation and workflow abuse. “Fully autonomous evasion networks sit on the horizon,” he said. “They are serious, but they are not yet the everyday baseline.”
This distinction matters, said Gogia: “If enterprises obsess over cinematic autonomous agent scenarios while leaving remote hiring, vendor onboarding, payment approvals, and document review full of holes, they will lose in the most prosaic way imaginable.”
The report, he said, also gets the “asymmetry” right. “Offensive actors can learn across the ecosystem,” he said. “They can scrape open information, reuse leaked records, study enforcement patterns, test onboarding forms, inspect public procurement data, watch court filings, probe compliance thresholds and [use the information to] refine their behavior.”
Defenders, by contrast, are hemmed in by privacy rules, fragmented data, explainability requirements, jurisdictional boundaries, conservative operating models and siloed technology estates. “Offensive AI learns broadly,” he said. “Defensive AI often learns from fragments. That is the structural asymmetry.”
He explained that the regulatory landscape also amplifies the problem, in that regulatory bodies “still speak in separate dialects. [For example] the EU AI Act pushes organizations toward stronger obligations for high-risk AI. NIST-style frameworks push risk management, transparency, and governance.”
A trust architecture problem
Financial Action Task Force (FATF) expectations push national risk assessment and counter-proliferation controls, he noted, while banking regulators focus on model risk, accountability and operational resilience. “None of these streams is irrelevant. The trouble is that criminals do not organize themselves around regulatory workstreams. They organize around outcomes.”
What that means, said Gogia, “is that enterprise cannot wait for a clean global rulebook. It will not arrive in time. CIOs, CISOs, compliance officers and boards need a working governance model now. They need privacy-preserving analytics, controlled data environments, audit trails, legal safeguards and clear model-risk accountability.”
He said that enterprise IT managers should treat the situation as a trust architecture problem rather than a narrow sanctions-screening problem. “The uncomfortable truth is that AI is not simply helping bad actors write better phishing emails or forge tidier documents,” he noted. “It is helping them manufacture legitimacy across a chain of enterprise workflows.”
Likely outcome an ‘AI arms race’
Report author Arnold also noted that there are signs that cyber criminals have discovered new AI technologies and abilities that legitimate enterprises could adopt for legitimate applications.
History, he said, “is replete with [criminals] developing novel solutions to tough problems, [which are] later adopted by law enforcement. Much of our anti-financial crime policy is effectively a response to bad actors exploiting systems or using technology in novel ways to perpetrate crimes. In this scenario, I think an ‘AI arms race’ between enforcement authorities and bad actors is the most likely outcome.”
Gogia added, “the baddies are not teaching enterprises how to invent AI. They are teaching enterprises where trust is leaking. That is the lesson worth taking seriously.”
This article originally appeared on CIO.com.
View the full article
CSOonline

AI models more vulnerable than claimed when faced with iterative attacks

CISOs relying on LLM runtime guardrails and official safety scores when making security decisions about their organizations’ AI usage and model selection are due for a wakeup call.
According to a new study from Cisco, frontier models from OpenAI, Anthropic, Google, xAI, and Amazon have significantly worse risk profiles when pressured in multi-turn attacks compared to when their safety is benchmarked using single prompts.
“The dominant safety benchmarks for frontier large language models share a structural assumption: that a single prompt and a single model response are enough to characterize how a model behaves under adversarial attack,” the Cisco researchers who authored the study said in a blog post. “These benchmarks inform model cards, safety reports, and procurement decisions across the industry, but they all only measure one narrow slice of attacker behavior.”
Instead, the researchers subjected 15 of the most widely used frontier AI models to a variety of attack techniques that are more likely to occur in the real world, where attackers will not give up after the model refuses to respond to one malicious prompt.
“Real adversaries iterate,” the researchers said. “They reframe refusals, decompose tasks across turns, adopt personas, and escalate gradually. A single turn benchmark cannot see any of that.”
Stress-testing over multiple prompts
The tests pitted various model configurations, such as with reasoning enabled or disabled, against a range of attack strategies aimed at bypassing safety guardrails. Techniques included role-play; misdirection or introducing ambiguity into the context; redirection or reframing the model’s refusal; information decomposition and reassembly; and incremental escalation, by breaking a task into smaller parts that don’t seem malicious on their own.
The researchers ran 30,090 single prompt attacks (2,006 per model) to determine the weighted single-turn attack success rate (ASR) for every model and then ran 6,986 multi-turn attacks across 1,456 conversations for comparison. The results were telling: Most models had considerably higher average ASR scores for multi-turn attacks compared to single-prompt attacks.
For example, Anthropic’s Claude Opus 4.6 and OpenAI’s GPT 5.4 — the latest versions at the time of testing — had single-turn ASRs of 3.64% and 2.74%, respectively. When faced with multi-turn attacks, average ASRs jumped to 16.20% for Opus and 24.68% for GPT.
Neither of those, however, represented the biggest score jump. Google’s Gemini 3 Pro had a single-turn ASR of 18.10% and a multi-turn ASR of 73.35%.
“For business decisions made on the basis of published single-turn scores, this presents security and governance risk,” the researchers concluded. “A model with 2.74% single-turn ASR is not the same product as a model that holds the line at 24.68% multi-turn ASR. Without paired-regime data, the two are indistinguishable on most public evaluations, and the end user never sees the gap.”
Cisco
The results also revealed that different model configurations can impact safety. For example, xAI’s Grok 4.1 Fast in non-reasoning mode had the worst multi-turn ASR at 88.30%, but its score dropped to 43.47% when reasoning was turned on. The researchers note that these configuration-related variations are not currently captured by the official model cards published by the labs or the public safety benchmarks.
Different attack strategies showed meaningful differences in success across models, both for single-turn and iterative attacks — findings that could be used to inform defense strategies for customers of these models.
The tests also uncovered outliers such as Amazon’s Nova Lite, Nova Lite 2, and Nova Micro models, all of which had more than three times higher single-turn ASRs than multi-turn ones.
Open-source models from labs such as Meta, Mistral, Alibaba, DeepSeek, Google, OpenAI, Zhipu, and Microsoft faced the same challenges when it came to multi-turn attacks, as highlighted in a study published in November by the same Cisco research team.
“Taken together, the two studies make a stronger claim than either alone: multi-turn vulnerability is a structural property of the current frontier, not an artifact of open-weight alignment choices or capability-first development,” the researchers said. “Whether the weights are public or proprietary, whether the lab prioritizes safety or capability, the iterative attack surface remains an open challenge across the frontier.”
Call to action
Cisco’s researchers are calling for better benchmarks that consider real-world attacks and AI-specific vulnerabilities as identified by OWASP and other organizations, instead of primarily focusing on content safety.
Model creators should also be more transparent about how various configuration flags — such as reasoning modes, temperature, and system prompt adherence settings — impact safety, according to the researchers. They should also publish ASRs for both single-turn and multi-turn attacks, further split across various attack strategies.
This is especially important given that upcoming regulatory frameworks such as the NIST AI Risk Management Framework, the draft NIST Cyber AI Profile (IR 8596), and Article 15 of the EU AI Act call for adversarial testing.
“Any model with an absolute gap >15 [percent points] between single-turn and multi-turn ASR should trigger a manual review before deployment,” the researchers said. “In this cohort that rule flags eight models: five with positive deltas (Gemini 3 Pro; Grok 4.1 Fast NR; GPT-5.4; Grok 4.1 Fast R; GPT-5.2) and three with negative deltas (Nova Lite; Nova Micro; Nova 2 Lite).”
View the full article
CSOonline

Roku Unveils Redesigned Home Screen

Roku today announced that its home screen is receiving its biggest redesign in more than a decade. The new home screen will begin rolling out to Roku TVs and streaming devices across the U.S. starting today, with more countries to follow later. The changes will appear automatically, with no software update required.


At the top of the home screen, a new "Top Picks for You" section provides recommendations based on your watch history and what is trending on Roku.

A new AI-powered "Quick Access" section surfaces the apps you use most, such as the Apple TV app. There is still a separate page containing a complete list of apps that you have installed on your Roku device if you need it.


New curated hubs called "Destinations" offer content organized by genres and moods, such as comedy, reality, movies, sports, documentaries, and more.

The menu on the Roku home screen now has a collapsed design with icons, rather than expanded text labels. Roku says this change provides a cleaner canvas for discovering content, but it also results in a large ad being shown persistently on the right side of the screen. Previously, the ad would only appear in certain views.


A new "Shortcuts" section provides quick access to features like Continue Watching, Sleep Timer, Save List, and more, all in one place.

If you are a fan of the Roku City screen saver, there is now an on-screen tile that you can select to view it at any time. You can explore the screen saver's iconic cityscape, plus access Daily Trivia and other mini-games.Tags: Home Screen, Roku
This article, "Roku Unveils Redesigned Home Screen" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Developing iPhone Anti-Snatching Feature That Locks Stolen Phones Instantly

Apple is developing a new feature that will lock your iPhone if it's snatched from your hand by a thief, according to Apple code seen by 9to5Mac. The option will use the gyroscope, accelerometer, and other sensors to determine when an iPhone has been grabbed. It'll also rely on a paired Apple Watch to detect when the iPhone has suddenly moved away from the owner's wrist.


Once the iPhone is yanked from your hand, it will lock and activate Stolen Device Protection to prevent thieves from accessing information on it.

Stolen Device Protection adds extra security to your iPhone when you're away from familiar locations like home or work. It requires biometric authentication for actions like accessing stored passwords or credit cards, and there are built-in hour-long delays for actions like changing an Apple Account password.

The feature was originally designed to protect iPhone users from stealthy thieves who observe someone's passcode and then snatch an iPhone. With a passcode, thieves could get into apps and access bank account data and other sensitive information, but Stolen Device Protection prevents that from happening.

Android already has a Theft Detection Lock feature that locks a smartphone in a snatch-and-grab theft situation.

There is no word on when the new feature might be added to the iPhone.
This article, "Apple Developing iPhone Anti-Snatching Feature That Locks Stolen Phones Instantly" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Meta Wants You to Pay for Instagram, Facebook, and WhatsApp Now

Meta is rolling out paid Instagram Plus, Facebook Plus, and WhatsApp Plus plans worldwide as of today.


Instagram Plus is priced at $3.99 per month, Facebook Plus is priced at $3.99 per month, and WhatsApp Plus is priced at $2.99 per month. According to TechCrunch, the paid plans provide features like profile customization, super reactions, and story insights.

Instagram Plus lets users see how many people have rewatched a Story, and it adds unlimited audience lists for Stories for creating groups other than Close Friends. Users can spotlight a story once a week for extra views, use Super Heart animated reactions, choose custom app icons, add customized fonts to a profile bio, extend a story beyond 24 hours, and search a story viewer list to see who is watching. Subscribers will also be able to post straight to their profiles without having the post show up in their followers' feeds, and they will be able to stealthily "preview" Instagram stories without showing up as a viewer.

Facebook Plus includes most of the same features as Instagram Plus, while WhatsApp Plus includes app themes, custom ringtones, more pinned chats, list customization, and premium stickers.

Meta head of product Naomi Gleit said the company is also exploring new subscription plans for creators and businesses, along with plans for AI users. The new plans are being offered under "Meta One" branding that combines subscription offerings from multiple Meta platforms.

The $7.99 Meta One Plus plan and the $19.99 Meta One Premium plan are aimed at Meta AI users. Both plans unlock higher compute queries, reasoning, and image/video generation, but Premium offers more capacity, including deeper reasoning for complex tasks.

A Meta One Essential plan priced at $14.99 per month is designed for creators and businesses. It includes a verified badge, impersonation protection, better analytics, and a linksheet that lets users link to their online profiles on the web and on other social media networks. The $49.99 Meta One Advanced plan includes the Essential options plus features in the Facebook feed, optimized scheduling tools, notifications when others reuse a creator's content, higher rankings in Instagram and Facebook search, a bolder Follow button on Reels, and automatic follow invitations for people who engage with a creator or brand's content.

Meta is going to start testing the AI Meta One plans in Singapore, Guatemala, and Bolivia next month. The business plans will be tested in Saudi Arabia, Morocco, Thailand, and Bangladesh starting later this week.

Gleit described Meta One as a place that brings Meta subscriptions "together" across all Meta apps. She said Meta's new plans were "just the beginning with a lot more value to come."Tags: Facebook, Instagram, Meta, WhatsApp
This article, "Meta Wants You to Pay for Instagram, Facebook, and WhatsApp Now" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Samsung's 2026 Monitors Launch With $50 Discounts and Credit Towards Future Purchases

Samsung's newest monitors are now available to purchase this week, including the Odyssey G8, ViewFinity S8, and Movingstyle Essential. All of these are available with a $50 launch discount, plus your choice of extras including up to $300 in Samsung credit on a future purchase, a free Music Studio speaker, or free Galaxy Buds4 Pro.

Note: MacRumors is an affiliate partner with Samsung. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

Starting with the Samsung Odyssey G8 monitors, you will find $50 discounts across every model of the new 2026 devices. Prices start at $899.99 for the 27-inch Odyssey G8 5K Monitor, and also include Samsung's first 6K monitor with the 32-inch Odyssey G8 5K Monitor for $1,549.99.

$50 OFF PLUS EXTRASSamsung 2026 Monitors

Samsung also has a new 40-inch ViewFinity S8 Curved Monitor on sale for $1,349.99 for launch week, as well as the Movingstyle Essential Monitor for $849.99, both $50 discounts. Additionally, the company announced a 27-inch model of the ViewFinity S8, but it's not yet available for purchase.

43-inch Movingstyle Essential - $849.99 ($50 off), plus $200 Samsung credit
27-inch Odyssey G8 5K Monitor - $899.99 ($50 off), plus $200 Samsung credit
32-inch Odyssey G7 4K Monitor - $1,049.99 ($50 off), plus $200 Samsung credit
27-inch Odyssey OLED G8 4K Monitor - $1,049.99 ($50 off), plus $200 Samsung credit
32-inch Odyssey OLED G8 4K Monitor - $1,249.99 ($50 off), plus $300 Samsung credit
40-inch ViewFinity S8 Curved Monitor - $1,349.99 ($50 off), plus $300 Samsung credit
32-inch Odyssey G8 6K Monitor - $1,549.99 ($50 off), plus $300 Samsung credit

If you're on the hunt for more discounts, be sure to visit our Apple Deals roundup where we recap the best Apple-related bargains of the past week.



Deals Newsletter

Interested in hearing more about the best deals you can find in 2026? Sign up for our Deals Newsletter and we'll keep you updated so you don't miss the biggest deals of the season!




Related Roundup: Apple Deals
This article, "Samsung's 2026 Monitors Launch With $50 Discounts and Credit Towards Future Purchases" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Just Expanded iPhone Driver's License Feature to 14th U.S. State

In select U.S. states, residents can add their driver's license or state ID to the Apple Wallet app on the iPhone and Apple Watch, and then use it to display proof of identity or age at select airports and businesses, and in select apps.


Starting today, the feature is available in Arkansas, which is the 14th state to offer it. However, it may take some time to roll out to all users.

To set up the feature, open the Wallet app on the iPhone and tap on the plus sign in the top-right corner. Next, tap on Driver's License and ID Cards, select Arkansas from the list, and follow the on-screen steps to complete the process.

Supported States

The following states offer driver's licenses in the Wallet app:Arizona (since March 2022)
Maryland (since May 2022)
Colorado (since November 2022)
Georgia (since May 2023)
Ohio (since July 2024)
Hawaii (since August 2024)
California (since September 2024)
Iowa (since October 2024)
New Mexico (since December 2024)Montana (since August 2025)
North Dakota (since September 2025)
West Virginia (since October 2025)
Illinois (since November 2025)
Arkansas (since May 2026)The feature is also available in Puerto Rico.

Future States

Apple or local DMV offices previously announced that the following U.S. states had signed on to adopt the feature, but no timeframes were disclosed:Connecticut
Kentucky
Mississippi
Oklahoma
Utah
VirginiaParticipating Airports

Apple Wallet IDs are accepted at TSA checkpoints in more than 250 airports in the U.S., for domestic travel. Given that Apple Wallet IDs are not accepted by law enforcement, and lack many other use cases, carrying a physical ID is still necessary.

Here are just some of the airports that offer the feature — there are hundreds of others:Baltimore/Washington International Thurgood Marshall Airport (BWI)
Ronald Reagan Washington National Airport (DCA)
Washington Dulles International Airport (IAD)
Phoenix Sky Harbor International Airport (PHX)
Denver International Airport (DEN)
Hartsfield-Jackson Atlanta International Airport (ATL)
Cincinnati/Northern Kentucky Airport (CVG)
John Glenn Columbus International Airport (CMH)
San Francisco International Airport (SFO)
San Jose Mineta International Airport (SJC)
Los Angeles International (LAX)
Daniel K. Inouye International Airport (HNL)
Des Moines International Airport (DSM)
Eastern Iowa Airport (CID)
Albuquerque International Sunport (ABQ)
Lea County Regional Airport (HOB)
Luis Munoz Marin International Airport (SJU)
Billings Logan International Airport (BIL)
Bozeman Yellowstone International Airport (BZN)
Great Falls International Airport (GTF)
Missoula International Airport (MSO)
Chicago O'Hare International Airport (ORD)
Chicago Midway International Airport (MDW)
St. Louis Lambert International Airport (STL)
Richmond International Airport (RIC)
Norfolk International Airport (ORF)
Roanoke-Blacksburg Regional Airport (ROA)
Newport News-Williamsburg International Airport (PHF)Travelers should refer to TSA signage to confirm availability of the feature.

Digital Passport


If you live in a state that does not yet offer Apple Wallet IDs, you can now create a Digital ID based on your U.S. passport, and present it at the same participating TSA checkpoints, for age and identity verification purposes during domestic travel. It is not a replacement for a physical passport, and it cannot be used for international travel.

This feature requires iOS 26.1 or watchOS 26.1 and later.Tag: Apple Wallet
This article, "Apple Just Expanded iPhone Driver's License Feature to 14th U.S. State" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Atlanta's MARTA Breeze Card Now Available in Apple Wallet

Starting today, the Metropolitan Atlanta Rapid Transit Authority (MARTA)'s Breeze Card can be added to the Apple Wallet app, allowing commuters in the Atlanta, Georgia area to tap an iPhone or Apple Watch to pay their fare at train stations and on buses.


To add the Breeze Card to Apple Wallet, open the app and tap on the plus sign in the top-right corner. Next, tap on Transit Card and scroll down and select the Breeze Card. Then, follow the on-screen steps to complete the setup process.

Funds can be added to the Breeze Card directly in Apple Wallet, with both pay-per-ride and discounted passes for multiple rides available.

Express Mode allows commuters to simply hold their iPhone or Apple Watch near a payment reader to pay for their ride, without needing to unlock or wake the device. This mode even works for up to five hours after an iPhone runs out of battery. Express Transit can be enabled per card in the Settings app under Wallet & Apple Pay.

Breeze Card has been available in the Google Wallet and Samsung Wallet apps since earlier this month, but Apple Wallet support was just added now.Tags: Apple Wallet, Express Transit
This article, "Atlanta's MARTA Breeze Card Now Available in Apple Wallet" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

FastAPI-based AI tools exposed to authentication bypass by flaw in Starlette framework

A single malformed character in a web request can let an unauthenticated attacker slip past the access controls that guard applications built on Starlette, the open-source Python framework that powers FastAPI, researchers said.
The flaw, tracked as CVE-2026-48710 could allow attackers to bypass host-validation protections using malformed Host headers, according to an advisory from cybersecurity firm X41 D-Sec.
The attacker needs no password and no action from a victim, it said.
Starlette’s maintainer released a patch through an official GitHub security advisory after X41 D-Sec disclosed the vulnerability in coordination with the Open Source Technology Improvement Fund (OSTIF). They found the flaw during an unrelated source-code audit, and traced it to Starlette rather than the application under review.
“This bug is a classic ‘responsibility gap’ where if this maintainer didn’t patch, thousands of exposed projects would have to individually secure their projects,” OSTIF said.
The researchers have created a website, badhost.org, that can test websites for the vulnerability.
Exploiting the bug
The flaw lies in how Starlette rebuilds the address of an incoming request, according to X41 D-Sec. The framework joins the Host header sent by the client to the path that was requested to form a complete URL, but parses the whole and the parts for validity using different rules.
A Host header containing a slash, question mark or hash character shifts where the path begins, the researchers said, so the path Starlette reports no longer matches the one the server actually received.
That gap is where the risk lies, according to the firm. Starlette routes the request to the real path, but middleware and endpoints read the altered one. An application that restricts sensitive routes by checking the path it sees can let a request through while still running the protected route behind it.
X41 D-Sec published a demonstration with its advisory. The researchers sent a request to a protected administrative page and received a “403 Forbidden” response. They sent the same request with one extra character in the Host header, and the page returned a “200 OK.” The same pattern has surfaced in other recent authentication-bypass flaws in open-source AI frameworks.
Severity rating under dispute
Starlette’s maintainer rated the flaw at 6.5 out of 10, or Moderate, on the CVSS scale in the GitHub advisory. X41 D-Sec rated it 7.0, or High, and said the danger to software built on Starlette runs higher than either figure suggests.
The damage an attacker can do depends on what each application does with the forged path. X41 D-Sec said it found several open-source projects whose security checks rely on the reconstructed address. In those projects, the single-character flaw could chain into “authentication bypass to SSRF and other issues that in some cases even lead to remote-code-execution on the affected system,” the researchers wrote.
The reach extends well past Starlette itself. A separate advisory from security firm Secwest on the flaw said the score “materially understates the downstream impact” and warned that the bug touches “most of the model-serving, gateway, proxy, eval, agent, and MCP-server infrastructure that has been stood up in the last two years.”
Affected software includes model-serving tools, API gateways, OpenAI-compatible proxies, agent frameworks and Model Context Protocol servers built on FastAPI, according to X41 D-Sec and Secwest.
An application can be exposed even if its developers never installed Starlette, because another component may have, X41 D-Sec said. Starlette has more than 400,000 dependent projects on GitHub, according to the firm.
Who is most exposed
Not every dependent project is equally at risk, X41 D-Sec said. Whether an application can be attacked comes down to how it is. The dividing line is the reverse proxy: A proxy such as nginx or Apache HTTP Server rejects the malformed request before it reaches the application, and production websites usually sit behind such a layer. Research, evaluation and development setups for AI software often do not, and many run the application server facing the network directly, it said.
Three groups face the most exposure, according to X41 D-Sec: those running a FastAPI or Starlette application directly on an application server with no compliant reverse proxy in front; those exposing a model proxy such as LiteLLM or vLLM as a directly reachable endpoint; and those whose access-control code reads the reconstructed request address rather than the raw path.
The researchers advised teams to upgrade to Starlette 1.0.1 or later, which validates the Host header and rejects malformed values.
View the full article
CSOonline

First Cases for Apple's Foldable iPhone Surface Online

Accessory maker iFunSmart has begun listing the first protective cases for Apple's upcoming foldable iPhone, corroborating rumors about the device's design.


Case makers routinely begin mass producing accessories ahead of a new iPhone announcement, working from dummy units or leaked CAD files to size their molds. Their designs are speculative, but they have historically proven accurate to the millimeter, since accessory makers cannot afford to be left without product on launch day. Leaker Sonny Dickson shared images of foldable iPhone dummy units in April, providing the kind of reference template that typically circulates among case manufacturers.

The listings, spotted by French Apple site iPhoneSoft, show an unobtrusive rear camera plateau housing two lenses, a slim profile, and a circular cutout for MagSafe-style magnets. iFunSmart describes the cases as offering military-grade drop protection, integrated N52 magnets, a translucent matte finish, and 1.5mm raised camera lips alongside a 1mm raised screen bezel.

The design broadly corroborates the design outlined in rumors and seen on dummy units, suggesting the foldable's exterior is increasingly clear. Only two camera lens cutouts are present, in line with reports that Apple plans to skip the telephoto camera. A cutout for a Camera Control button is also visible, but there is no Action button. The listing depicts a multi-part case with separate snap-on sections rather than a single-piece shell owing to its folding design.


The presence of magnets in the case does not necessarily mean Apple has built MagSafe into the foldable iPhone itself, and there has previously been speculation that the device could lack the feature. The N52 magnets could simply be embedded into the case to attach to external ‌MagSafe‌ accessories such as wireless chargers and car mounts, without aligning with a corresponding magnet array inside the device.

iFunSmart's listings are likely to be among the first of many. Accessory makers typically flood the market with cases in the months ahead of a new iPhone launch, and further variants from competing brands should appear in the run-up to the device's announcement.

Apple is widely expected to launch the device alongside the iPhone 18 Pro and ‌iPhone 18 Pro‌ Max in fall 2026. Leaks point to a 5.5-inch external display, a 7.8-inch inner display, a folded thickness of around 9.5mm and a thickness of about to 4.5mm when open, the A20 Pro chip with 12GB of memory, dual 48-megapixel rear cameras, and Touch ID in the side button rather than Face ID. The device is expected to start at around $2,000.Related Roundup: iPhone FoldTag: Foldable iPhone
This article, "First Cases for Apple's Foldable iPhone Surface Online" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

iPhone Air MagSafe Battery is Just $59

Amazon is still offering the iPhone Air MagSafe Battery for $59.00, down from $99.00. We started tracking this deal earlier in the month, and it beats the previous low price by about $20.

Note: MacRumors is an affiliate partner with Amazon. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

The iPhone Air MagSafe Battery is only compatible with the iPhone Air, and it can add up to 65 percent additional charge to the smartphone. The MagSafe Battery supports up to 12W of fast wireless charging, and it sports a thin and light design similar to the iPhone Air.

$40 OFFiPhone Air MagSafe Battery for $59.00

If you're on the hunt for more discounts, be sure to visit our Apple Deals roundup where we recap the best Apple-related bargains of the past week.



Deals Newsletter

Interested in hearing more about the best deals you can find in 2026? Sign up for our Deals Newsletter and we'll keep you updated so you don't miss the biggest deals of the season!




Related Roundup: Apple Deals
This article, "iPhone Air MagSafe Battery is Just $59" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Mitigating CVE-2026-31431 (“Copy Fail”) in Docker Engine

CVE-2026-31431 is a Linux kernel vulnerability that was recently disclosed. This CVE does not compromise Docker infrastructure.
That said, Docker Engine’s default profiles prior to v29.4.3 allowed containers to create AF_ALG sockets, which is the syscall surface the exploit uses. You are not exposed if you are running Docker Engine v29.4.3 or later, OR a patched host kernel. If either of those is missing, you have exposure on that host, and you should read the rest of this post.
As of writing, the kernel patch is available on Debian (CVE-2026-31431) and RHEL 9 (RHSB-2026-002) but not yet on Ubuntu. For users on distros that haven’t shipped a kernel fix, upgrading Docker Engine is the mitigation you can apply today.
Why you should read about Copy-Fail
This CVE drew a lot of attention because the exploit became public before many Linux distributions had kernel patches available. As a result, most distros were still vulnerable and had no ready fix at the time of disclosure. It was especially notable because the bug affected Linux kernels going back to around 2017, making the potential impact unusually broad.
On the Docker Engine team, I started investigating what we could do from our end to protect users on vulnerable hosts. It turned out the mitigation was more involved than it first looked, and the first attempt broke 32-bit binaries. This post is what we shipped, what broke, what we learned, and where things stand now.
What Copy Fail is
On April 29, researchers disclosed CVE-2026-31431, dubbed “Copy Fail,” a privilege escalation vulnerability in the Linux kernel’s AF_ALG crypto subsystem.
The flaw is in the algif_aead module. It allows any unprivileged user with access to an AF_ALG socket to perform controlled writes to the page cache. Since the page cache backs file reads across the entire system, an attacker can temporarily modify the contents of any readable file as seen by every process on the host. Corrupting a setuid binary is the most direct path to local root, but the primitive itself is more general.
The exploit is trivial and works on every unpatched Linux kernel shipped since 2017.
The correct fix is a kernel update. The mitigations described below reduce exposure for containers running on unpatched kernels, but they do not fix the underlying vulnerability. If your kernel vendor has released a patch, apply it.
What does this mean for containers?
Inside a container running with default security profiles, an attacker with code execution can use Copy Fail to corrupt pages in the page cache. One possible outcome is escalating to root inside the container by corrupting setuid binaries.
But the page cache is shared across the host, so the impact is not confined to the attacker’s container. Modified pages are visible to the host and to every other container that maps the same file, including shared image layers. Other workloads on the same node can be affected.
The attack does not require any special capabilities or privileges beyond what a default container provides. The only requirement is the ability to create an AF_ALG socket, which was previously allowed by Docker’s default security profiles.
First attempt: seccomp (v29.4.2)
We updated Docker Engine’s default seccomp profile to block AF_ALG sockets. The seccomp filter inspects the first argument to socket(2) and denies address families AF_ALG and AF_VSOCK (which was already blocked).
Blocking socket(2) is not enough on its own. There is another way to create sockets on x86_64 Linux: socketcall(2), an older multiplexed syscall that wraps socket, bind, connect, and other socket operations behind a single syscall number.
There is another way to create sockets on Linux: socketcall(2), an older multiplexed syscall that wraps socket, bind, connect, and other socket operations behind a single syscall number.
The problem for seccomp is that socketcall packs the real arguments (including the address family) into a userspace array and passes a pointer, which BPF cannot dereference and inspect. There is no way to selectively block AF_ALG through socketcall with seccomp.
Linux 4.3 already added direct socket syscalls for i386 and s390, so we assumed most modern binaries would already use the new socket syscall and that socketcall would only matter for old binaries. So we blocked it entirely and shipped Docker Engine v29.4.2 (release notes).
What broke
The socketcall deny turned out to be too broad.
Older versions of glibc on i386 route all socket operations through socketcall, the Go runtime uses it unconditionally for GOARCH=386 (independent of glibc), and many legacy and gaming workloads (SteamCMD, Wine) depend on it.
Blocking socketcall broke networking for a lot of 32-bit binaries running inside a container (moby/moby#52506).
And this is not just an i386 problem. On amd64, any process can switch into ia32 compatibility mode with int $0x80 and invoke socketcall directly, bypassing the socket(2) arg filter entirely. You do not need a 32-bit container or a 32-bit binary to reach that path.
Affected containers could work around this by using a custom seccomp profile that re-enables socketcall while keeping AF_ALG blocked for the direct socket(2) path.

But that just pokes a hole in the hardening for those containers, since an attacker inside them could still reach AF_ALG through socketcall.
Second attempt: LSM-based enforcement (v29.4.3)
The fundamental problem is that seccomp operates at the syscall boundary, and socketcall multiplexes many operations behind a single syscall number with pointer arguments. You cannot selectively block AF_ALG through socketcall with seccomp alone.
AppArmor and SELinux operate on a different level. Linux Security Modules hook directly into the kernel’s security_socket_create() callback, which fires when the kernel actually creates the socket object, regardless of which syscall entry point was used. An LSM can deny AF_ALG specifically while leaving all other socketcall usage intact.
In v29.4.3 (release notes), we:
Reverted the socketcall seccomp deny to restore 32-bit compatibility. Added deny network alg, to the default AppArmor profile (moby/profiles#22).
On systems with AppArmor enabled (e.g. Ubuntu, Debian), this blocks AF_ALG through both socket(2) and socketcall(2). Integrated a SELinux CIL policy module for systems running SELinux (Fedora, RHEL, CentOS).
The module denies alg_socket creation for all container_domain types and can be loaded via semodule.
SELinux enforcement requires the daemon to be running with --selinux-enabled. Kept the seccomp socket(AF_ALG) arg filter as defense-in-depth for the direct socket(2) syscall path. What you should do
Patch your kernel.
This is the real fix.
Check with your distribution for a kernel update that addresses CVE-2026-31431. Upgrade Docker Engine to v29.4.3 or later. You get the updated seccomp + AppArmor + SELinux defaults. A systemctl restart docker (or equivalent) is enough; no host reboot required. If you cannot upgrade the kernel or the engine immediately: Blacklist the kernel modules: add blacklist af_alg and blacklist algif_aead to /etc/modprobe.d/.
This only works if the modules are built as loadable modules (CONFIG_CRYPTO_USER_API=m), not compiled into the kernel. Apply a custom seccomp profile that denies AF_ALG using --security-opt seccomp=/path/to/profile.json or the seccomp-profile option in daemon.json. Closing thoughts
Security comes in layers, and sometimes no single layer is enough. Seccomp blocks socket(AF_ALG) on every system but is blind to socketcall. AppArmor and SELinux block both paths, but they depend on host configuration. Together, they cover what neither can alone.
On systems without an LSM, the socketcall path remains unblocked from Docker’s side. Ultimately, the kernel bug is what needs to be fixed.
Kernel vulnerabilities will keep coming. When they do, the container runtime is often the fastest place to deploy a mitigation, because updating the engine is one change that protects every container on the host. The Copy Fail timeline made that especially clear: the embargo broke before distros had fixes ready, and for several days the engine was the only place users could mitigate anything without waiting for a kernel rebuild.
Keeping Docker Engine up to date is not just about new features. It is one of the most effective ways to shrink the window between a kernel CVE going public and your workloads being protected against it.

View the full article
reporter

Apple Updates Trade-In Values for iPhone, iPad, Mac, and Apple Watch

Apple today updated its U.S. trade-in estimates, raising values for most current iPhone, iPad, Mac, and Apple Watch models while reducing several Android offers.


The headline iPhone trade-in figure rises from $685 to $695, with every iPhone 16 model gaining value:


iPhone 16 Pro Max: $685 to $695
iPhone 16 Pro: $550 to $560
iPhone 16 Plus: $455 to $465
iPhone 16: $435 to $460


Every current ‌iPad‌ also gains value, with the headline range shifting from $40 to $670 up to $45 to $690:


iPad Pro: $670 to $690
iPad Air: $445 to $460
iPad: $220 to $235
iPad mini: $250 to $265


Mac changes are mixed, with most current models gaining value:


MacBook Pro: $685 to $690
MacBook Air: $485 to $520
Mac mini: $340 to $375
iMac: Unchanged at $355


Despite those increases, the top of Apple's Mac trade-in range slips from $2,090 to $2,045, suggesting a reduction for a higher-end model not shown in Apple's summary table, such as the Mac Pro or Mac Studio. The Apple Watch lineup also gets a mix of revisions:


Apple Watch Ultra 2: $295 to $305
Apple Watch Series 9: $120 to $130
Apple Watch Series 10: Unchanged at $150
Apple Watch Ultra: $215 to $205


Android trade-in offers have largely been cut, with the headline range narrowing from $30 to $370 down to $30 to $360:


Samsung Galaxy S23 Ultra: $230 to $200
Google Pixel 8 Pro: $170 to $165
Samsung Galaxy S23: Unchanged at $125
OnePlus 12: Unchanged at $200


All Apple-listed values are estimates, with final offers determined after the device is received and inspected. Customers can apply trade-in credit toward a new purchase or receive the value as an Apple Gift Card, and Apple will recycle ineligible devices for free.Related Roundup: Apple Watch 11Tag: Apple Trade-InBuyer's Guide: Apple Watch (Caution)
This article, "Apple Updates Trade-In Values for iPhone, iPad, Mac, and Apple Watch" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple's 'After the Whistle' Podcast to Return for World Cup

Apple today announced that "After the Whistle with Brendan Hunt and Rebecca Lowe" will return on June 7 for a third season built around the 2026 FIFA World Cup.


Hunt is an actor and cocreator of Apple TV's Ted Lasso. Lowe hosts NBC Sports' Premier League coverage and is cohosting FOX Sports' FIFA World Cup 2026 broadcast. The two will recap games as the tournament unfolds, with new episodes landing multiple times a week in the hours after notable matches.

The show is produced by Apple News and presented by Verizon, and will be available in audio and video on ‌Apple News‌, Apple Podcasts, and other podcast platforms. The first episode arrives on June 7 with tournament previews.

Alongside the podcast, the ‌Apple News‌ app will feature tournament coverage from outside publishers, the schedule, scores, brackets, and player feeds. The free Apple Sports app, which Apple expanded to 90 more countries earlier this month, will offer live scores, stats, and a bracket view for the tournament.Tags: Apple News, Apple Podcasts, Apple Sports
This article, "Apple's 'After the Whistle' Podcast to Return for World Cup" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Publishes Document to Help Users Tell Creator Studio Apps Apart

Apple yesterday published a support document to help users distinguish Apple Creator Studio versions of its professional creative apps from the standalone editions sold as one-time purchases.


The confusion stems from Apple's decision to ship two parallel variants of Final Cut Pro, Logic Pro, MainStage, Motion, Compressor, and Pixelmator Pro, with one available through the Apple Creator Studio subscription and one sold separately. Both editions share the same name and can be installed on the same Mac at the same time, leaving little to tell them apart at a glance.

Apple's solution is to give the Creator Studio versions of the apps redesigned icons with Liquid Glass. The new support document presents side-by-side icon comparisons for each of the six apps so users can identify which edition they are running or troubleshooting from the Dock or the Applications folder.

Apple does not typically publish a dedicated reference document for telling two of its own apps apart, and the move suggests the dual-version setup has produced enough real-world confusion to warrant public guidance.

Apple Creator Studio launched in January for $12.99 per month or $129 per year, bringing the company's pro creative apps under a single subscription. Apple said that some new features in its creative apps would be available only to subscribers going forward. Pixelmator Pro's inclusion was the first significant sign of how Apple is integrating the Pixelmator team, which it acquired in November 2024.Tag: Apple Creator Studio
This article, "Apple Publishes Document to Help Users Tell Creator Studio Apps Apart" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

The NSA, ‘Mythos’ and the quiet emergence of AI cyber doctrine

For most of my career running security operations, the shape of cyber conflict has been defined by who could move faster than the other side. Faster at identifying a vulnerability, faster at patching, faster at detecting, faster at responding. The last few months have made me reevaluate that framing. Speed still matters. It just no longer carries the picture on its own. Scale and autonomy have moved alongside it, and the relative emphasis I place on the three is something I expect to keep adjusting. When I read recent coverage of the US government’s deepening use of advanced AI for cyber operations, Anthropic’s Claude Mythos Preview disclosure and the wave of defensive AI being built in response, I recognized the pattern. It fits the pattern of doctrine forming.
Doctrine rarely arrives through formal announcements in this field. It emerges through repeated behavior, through choices made under operational pressure, through what capable actors do when no one is telling them to stop. That is where I believe we are now.
From tools to operational capability
I remember when cyber operations lived inside scripts. They moved into frameworks, then into automated pipelines, then into what we somewhat optimistically called orchestration. Each step compressed time and lowered required expertise. Frontier AI is starting to look to me less like the next step in that sequence and more like a different thing.
What seems to separate frontier AI from the automation we have lived with, in what I have seen so far, is less about efficiency and more about independence. A model that can conduct reconnaissance across an unbounded attack surface, identify vulnerabilities without predefined signatures, assist in exploit chaining and adapt based on feedback feels less like enhancing an analyst’s workflow and more like operating with reduced human constraint. That shifts the economics of offense in ways that break assumptions most security programs still quietly rely on.
The Mythos Preview disclosure made the shift concrete. The model reportedly surfaced thousands of high-severity vulnerabilities, including findings in every major operating system and web browser, and chained multiple vulnerabilities into novel attacks with limited human direction. A specific example that landed for many readers was a 17-year-old remote code execution flaw in the FreeBSD NFS server (CVE-2026-4747), which Mythos identified and exploited autonomously after a single prompt. The defensive coalition Anthropic assembled under Project Glasswing includes AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA and Palo Alto Networks, with extended access reaching more than forty additional organizations responsible for critical software infrastructure, backed by roughly $100M in usage credits and $4M in donations to open-source security work. That is not a marketing exercise. It is a coordinated reaction to a threat model that has already moved. The fact that the coalition is now drawing antitrust scrutiny is itself a signal: This is no longer experimental.
The line that stayed with me from Anthropic’s own writeup was that the model could execute multi-stage attacks on vulnerable networks and discover and exploit vulnerabilities autonomously, completing in hours what would take human professionals days. Pair that with multiple frontier models from OpenAI now operating at the “High” cybersecurity threshold under its Preparedness Framework, including a defender-permissive variant (5.4-Cyber) built specifically for verified security teams, and with the disclosed incident of GTG-1002, the Chinese state-sponsored actor Anthropic publicly attributed in November 2025, jailbreaking Claude Code (by fragmenting tasks and posing as a defensive testing employee at a legitimate cybersecurity firm) to automate 80 to 90 percent of an operation that touched roughly 30 global targets and successfully breached four, and the trajectory stops being speculative. It is observable. The November 2025 GTG-1002 disclosure already touched regulated sectors, including financial institutions and chemical manufacturing, and AI-assisted pre-positioning against critical infrastructure is now documented in nation-state activity reports. The named, attributed, high-impact incident that will make this concrete to a board has not yet occurred publicly. The pattern is no longer hypothetical.
The doctrine forming in plain sight
Policy frameworks are still catching up. Reporting from Defense One over the past weeks makes clear that the US government is actively procuring AI-enabled vulnerability scanning, exploit development, threat data analysis and covert cyber infrastructure. The signal has now moved from procurement to codified policy: The FY 2026 NDAA directs the Department of Defense to develop an AI cybersecurity framework and incorporate it into DFARS and the CMMC program. Former senior NSA voices are discussing openly how AI reshapes offensive operations. The White House cyber posture has shifted toward more explicit offense, and that posture is being matched by capability. The experimental phase is over. We are in the operational one.
When a state-level actor integrates a new class of capability into live operations, doctrine follows. It does not get announced. It gets revealed through what targets are hit, how fast, at what scale, with what level of human oversight. The early outlines of AI cyber doctrine are already visible if you read the signals together.
Speed over stealth is the first. In an environment where exploit windows compress from weeks to hours, operating faster than a defender can respond is often more valuable than remaining undetected. That reverses the stealth-first operational model that shaped two decades of advanced persistent threat thinking.
Adaptive systems over static controls is the second. Playbooks that assume attacker behavior will repeat are already brittle. Phishing becomes dynamic. Malware morphs faster than signatures. Attack chains execute inside the time required to schedule an incident bridge. Defense either learns and adjusts, or it absorbs.
Probabilistic defense is the third. Zero-loss security was always a marketing ideal rather than an operational target, but the mismatch is now acute. The realistic objective is bounded loss: Assume continuous low-level compromise attempts are occurring, and optimize for detection, containment and minimized blast radius. I have had that conversation with peers more times in the last quarter than in the previous three years combined.
These are not constructs I am importing from a policy paper. They are the operational principles I see other security leaders quietly adopting because the environment does not offer another option.
Underneath those principles sits an economic shift I keep coming back to. Historically, attackers were constrained by three things: time, cost and expertise. AI compresses all three simultaneously. The NCSC’s most recent analysis frames the shift in concrete terms: In early 2026, the best frontier model completed nearly six times more attack steps on a realistic simulated enterprise attack than the best model eighteen months earlier, and a full attempt now costs around £65. Reconnaissance is continuous rather than episodic. Vulnerability discovery scales beyond any human team. Attack generation is iterative and cheap. Defense, meanwhile, is still indexed to human speed and decision-making. Offense is operating at machine speed and scale, while defense is still paging analysts during incidents. That is the imbalance. What I’m seeing reads less like a tooling gap and more like a model mismatch.
The UK National Cyber Security Centre’s recent analysis of defensive advantage against frontier AI captured something I have struggled to articulate to my own executive stakeholders: defensive advantage is not a static condition. It has to be actively retained against a capability frontier that is moving faster than most governance structures can accommodate. Organizations that treat AI as an enhancement layer will be outpaced by organizations that treat it as a structural change to how security is designed.
What I think leaders should actually do
Three things, and I do not consider any of them optional.
1. Treat AI agents as security principals
Any autonomous or semi-autonomous AI system with access to sensitive systems, data or workflows needs the governance posture applied to privileged users. Identity, access control, behavior monitoring, audit. If an AI agent can act, it can cause harm, and it has to be governed accordingly. Calling it a tool absolves no one, and the scariest version of this problem is an internally sanctioned AI agent with broad access that nobody has scoped as a principal. This recommendation is no longer outside the consensus. NIST’s Center for AI Standards and Innovation formally launched the AI Agent Standards Initiative in February 2026, the NCCoE has issued a concept paper on software and AI agent identity and authorization, and identity vendors, including Okta, Microsoft and Google, have shipped first-class agent identity primitives. The line is drawn. The question is whether you cross it now or after an incident forces it.
2. Invest in adaptive defense rather than incremental detection
Adding another static-signature layer to an environment where attackers iterate at machine speed is mostly theater at this point. The investment that produces compounding returns is in defenses that learn, including the capacity to run AI-driven detection and response inside the seams where human review cycles used to live. That requires hard choices about where to reduce analyst toil, where to accept probabilistic outputs and where human judgment is still the right bottleneck.
3. Reframe the risk model
Build the program on the assumption that continuous low-level compromise attempts are the normal operating condition rather than the exception. The rare high-impact event framing is a residue of a threat environment we no longer live in. Budgets, metrics and executive conversations should reflect that shift. Board reporting built around annualized loss expectancy will not survive contact with an adversary operating on hour-long cycles.
For years I told my teams that the advantage in cyber went to whoever had the better tools. I was wrong, or at least incomplete. The advantage now goes to whoever adapts faster. Governments are already integrating these capabilities into live operations. The doctrine is not coming. It is forming, quietly, operationally and in plain sight. The question is whether defenders will recognize it in time to shape their side of it.
This article is published as part of the Foundry Expert Contributor Network.
Want to join?
View the full article
CSOonline

DSPM buyer’s guide: Top 10 data security posture management tools

Data security posture management (DSPM) explained
Data security posture management (DSPM) tools help security teams examine their entire data environment to find shadow data, reducing the risk of data loss.
Tracking down sensitive data across both cloud and on-premises systems can be vexing. Each environment presents its own challenges. Given the dynamic and ephemeral nature of cloud computing, cloud data is easily created, deleted, or moved around. The cloud attack surface is equally dynamic, making protection all the more difficult. On-premises data can be elusive, particularly when shadow AI usage creates mission-critical data stores outside IT’s purview. To address this latter point, most DSPM vendors are incorporating their own AI routines (or offering a separate AI SPM product, as we outlined here).
But AI isn’t the only source of shadow data: for example, old data repositories left lurking on some cloud container or in-house server that has long been forgotten, not updated, or unaccounted for. The goal of DSPM products is to locate this shadow data and complement the more expansive cloud security posture management (CSPM) tools. But instead of focusing on protecting cloud infrastructures, DSPM tools focus exclusively on the role of data and how it is consumed by various cloud and on-premises services.
How the DSPM market is evolving
Over the past few years, DSPM tools have been developed to discover both known and unknown data, provide structure, and manage the security and privacy risks of potential data exposure. The market segment has seen a lot of mergers and acquisitions of late, including the following:
Tenable acquired Eureka Security and Vulcan Cyber, folding them into its CNAPP Cloud Security platform Palo Alto Networks acquired Dig Rubrik acquired Laminar Security Proofpoint acquired Normalyze IBM acquired Polar Security, folding it into Guardium Veeam acquired Securiti Varonis acquired a variety of companies, including Cyral, SlashNext, and AllTrue.ai to enhance its DSPM and other products Thales acquired Imperva and created its CipherTrust DSPM Google acquired Wiz This activity shows how DSPM has become a hot commodity, as established security vendors are buying up niche vendors to expand how they identify and protect data.
According to Gartner, DSPM tools bridge “the gap between data discovery/classification and the eventual implementation of automated remediation controls.”
The research firm outlines five use cases for DSPM: data loss prevention (DLP), privacy and data governance, entitlement management, cloud posture management integration, and protecting AI-related workflows. DSPM has evolved to the point where “the products’ combined capabilities now bear only a slight resemblance to the capabilities of the vendors that pioneered DSPM, leading to uncertainty for customers as to what DSPM actually is,” Gartner says.
As a result, while all five use cases are key components of any DSPM strategy, not every DSPM tool will perform equally well — or even comprehensively — across the board.
Take DLP for example. It used to be the sole place where security tools would examine threats and try to combat risk from these threats. But as cloud estates have grown, and as AI training data exploded, enterprises need to move beyond plain-vanilla DLP and establish better ways to find evidence of an attack, stolen data, or AI-fueled phishing attempts.
Part of the problem is that, as Gartner reports, “traditional data security products have an insufficient view to discover previously unknown, undiscovered, or unidentified data repositories, and they fail to consistently discover sensitive data.” That is where integration into other security tools is essential — and why so many mergers and acquisitions have taken place to fill these gaps.
Another issue is that data usage can be messy: Many businesses have numerous data and application silos that don’t put data protection front and center. Moreover, organizations often lack consistent protections as their data spreads across clouds and applications — especially when it comes to finding its way into AI agents.
DSPM provides the locator function for these complex environments. Fixing the problems a DSPM tool finds is really the province of a various other security tools, such as security orchestration, automation, and response (SOAR); security information event management (SIEM); cloud-native application protection platforms (CNAPPs); and the like. Some DSPM vendors integrate or incorporate these “fix-it” tools with their products.
Overall, the DSPM market is experiencing a boom in interest, with the tools catching on quickly in the past few years. As late as 2022, Gartner found a miniscule market penetration of less than 1% across its clientele. More recently, the research firm found that DSPM’s growth over the past two years has outpaced every other cybersecurity category.
What to look for in data security posture management (DSPM) tools
DSPM tools require a significant amount of staffing resources to evaluate because they touch on so many aspects of your IT infrastructure. That’s a good thing, because you want these tools to seek out and find data no matter where it could be hiding.
Having a plan that prioritizes which data is most important to your organization will help focus your evaluation. It’s also important to document how each DSPM offering creates its data map and subsequent dashboards. You should also understand the specific cloud and on-premises services that are covered and which ones are on the vendor’s near-term product roadmap.
How each vendor describes where it goes looking for data is instructive. Every vendor supports some visibility into some of the cloud data repositories of Amazon Web Services, Google Cloud, and Microsoft Azure. But that doesn’t mean that they cover every service offered by each of the cloud providers that deals with data.
For example, AWS has its S3 storage, Relational Database Service, Redshift’s cloud data warehouse, Athena serverless SQL queries, and ElasticSearch managed data services, among several other places that operate on data. Veeam takes pains to delineate which services are covered in each cloud platform, but other DSPM providers are not as transparent. Varonis, meanwhile, uses a “universal data connector” that can seek out a wider range of structured data destinations, both cloud-based and on-premises. Some vendors acknowledge cloud services they don’t support. Be sure to note that this is a very dynamic situation as vendors are adding coverage areas continually as their customers demand them.
Tracking down data is just the beginning, however. Once found, the data must be cataloged, evaluated, and summarized in various dashboards. That could be tricky if done without tight security controls, which is why most DSPM vendors claim that “customer data always stays within the customer’s environment.” This typically means collecting metadata, rather than the data itself, using read-only access to the apps, services, and database structures. Vendors refer to this as “agentless” or “using API access.” This approach has the advantage of being able to scan huge volumes of data quickly to understand the nature of its usage and potential risk factors.
Once the data has been discovered and its metadata has been collected, the next step is to perform regular scans to see what changes have been made: Has data been copied to some dark corner of your cloud estate? Has someone just changed access rights to allow for greater or insecure access? These tools provide a single point of view across all the various cloud and on-premises data locations. The key word here is “regular.” Scans have default periods (such as daily or weekly) and can be activated when new data repositories are found.
How data is consumed in your production environment, including data pipelines, lakes, and warehouses, is another aspect to consider. This can involve creating data maps to classify this landscape as well as facilitating audits to enumerate who has access to which data resource and under what specific circumstances it was shared across your enterprise. Maps are not just pretty pictures but important visualizations that often show where shadow data was abandoned, for example.
On top of all these activities there is the entire field of data governance. DSPM products assign risks and apply consistent security policies to manage your entire data collection, and they work with other security tools to enforce these policies and remediate problems. Many of the vendors included below have begun offering continuous auditing as part of their DSPM governance package, which is a welcome development in this fast-changing world.
Each DSPM tool has several components, including agents and agentless collectors (useful for tracking on-premises data), a centralized management dashboard, scanners that detect and prioritize data collections, maps of data lineage and usage, and compliance assessments.
Most vendors offer their DSPM product in one or both wider contexts: to integrate with third-party security services (such as offered by Veeam and Wiz) or as part of their own security product portfolio with other add-on modules that include identity management, cloud management, detection and response, and log analysis tools (Palo Alto Networks, Varonis, and Wiz).
The specifics on these integrations are worth examining, as some vendors, such as Varonis, Wiz and Palo Alto Networks, have wider support than others. Understanding the scope, integration level, and what other protective features are included, and which are available at an extra cost will take some effort to figure out.
Products can be deployed as a complete SaaS cloud-based solution, run from on-premises servers or private virtual machines/containers, or in some combination.
Finally, there is the issue of pricing. All the products overviewed here are pricey. Few vendors were willing to share this information, indicating that prices are flexible and depend on numerous factors. Some vendors are now charging by the terabyte for DSPM, an interesting development that could result in higher usage costs. However, numerous vendors offer annual subscriptions on either or both the Amazon and Azure marketplaces, which typically start at $30,000 for the smallest of networks. Plan on spending at least $100,000 annually, with higher prices to analyze larger data collections.
Leading vendors for data security posture management (DSPM)
The market space of DSPM is evolving quickly. Based on our own research and research from Gartner, GigaOm, IDC, and other analyst firms, we’ve identified 10 DSPM tool providers worth investigating. We also contacted several other vendors for this article that did not respond to our inquiries, so they are not detailed here: BigID, Concentric, Flow Security, IBM, OneTrust, Rubrik, Symmetry Systems, and Theom.
Cyera DSPM Platform
Cyera’s DSPM Platform helps organizations discover, classify, and secure sensitive data across cloud, SaaS, AI, and on-premises environments. The platform — along with its DDR companion DataWatcher — enables enterprises to control what data AI applications and agents can access, govern how that data is used, and reduce exposure risk across data at rest, in motion, and in use. Cyera provides agentless visibility into structured, semi-structured, and unstructured data, highly actionable risk and access intelligence dashboards, and more than 500 built-in data classifiers. The platform integrates with numerous security and data ecosystem tools, including Netskope, Splunk, Tines, Wiz, Collibra, DataHub, and Secoda, and supports on-prem/hybrid deployments for customers requiring in-environment scanning and regional data residency controls. Pricing on AWS for its Cloud Platform starts at $50,000 per year.
Microsoft Purview DPSM
Microsoft Purview DPSM is part of a larger data protection effort that is in preview form and will be rolled out in June at the company’s Build conference. It consolidates separate data protection tools to provide a single place to monitor and enforce security policies, and create and monitor security objectives. Purview DPSM will integrate with DSPM tools from Cyera, BigID, and OneTrust, and it uses Copilot and other AI agents to identify and protect your data collections. It replaces older DSPM tools that are now labelled “classic” versions.
Palo Alto Networks Cortex Cloud DSPM
Palo Alto Networks’ Cortex Cloud DSPM integrates with hundreds of SIEM, workflow, and ticketing solutions, as well as XDR and single sign-on (SSO). It comes with more than 600 prebuilt data classifiers, but more importantly, it works closely with various AI tooling to automate detection and remediation threats. It supports Microsoft 365, Snowflake, other SaaS services, a wide range of cloud providers, and on-premises file shares.
Proofpoint DSPM
Proofpoint DSPM recently folded in the company’s Normalyze acquisition and added integration with Proofpoint’s DLP solution. The tool scans cloud, SaaS, and on-premises data sources. It uses AI tools to identify attack paths and classify high-value data and auto-remediates when identifying misconfigurations. It integrates out-of-the-box with APIs for SOAR, third-party ticketing, and notification and automation platforms, including Atlassian Jira, ServiceNow, Microsoft Purview, and Slack. It offers more than 300 data classifiers and protects AI workflows.
Sentra End-to-End Data Security Platform
Sentra End-to-End Data Security Platform provides both continuous compliance and a superset of DLP policies, offering deep support for most cloud computing services along with support for containers, virtual machines, and on-premises data sources. It has its own data detection and response (DDR) tool for near real-time detection and a series of actionable dashboards. There are lots of integrations with data management (Coralogix, DataDog, and DataHub), email, ITSM (Jira, PagerDuty, and ServiceNow), CNAPP (Wiz), collaboration (Atlan, Azure Boards, Monday.com, Slack, and Teams), IAM (Active Directory and Okta), incident response (Seemplicity), SIEM (Splunk), and on-premises file shares. It also has powerful data classifiers that leverage built-in AI features to provide metadata enrichment and context, along with automated risk mediation. Sentra has four pricing tiers on AWS, starting at $50,000 per year.
Tenable One Cloud Exposure
Tenable One Cloud Exposure combines DSPM with threat detection and integrates with data lakes and warehouses such as Atlas, Salesforce, ServiceNow, Snowflake, and Jira-based ticketing systems. The tool continuously scans, categorizes, and remediates data across on-premises, cloud, and SaaS platforms to correlate identity, workload context, and potential threats. It also can dynamically verify and protect data that is externally reachable. Tenable has expanded its protective envelope to cover additional data types (such as secrets) and cloud providers.
Thales CipherTrust DSPM
Thales CipherTrust DSPM ties together visibility and remediation of your data located across both cloud and on-premises environments. It builds on its Imperva acquisition and can classify both structured and unstructured data and can map and protect data flows and encryption keys and secrets across the entire enterprise.
Varonis DSPM
Varonis has been in the data security business for more than a decade and covers both on-premises and cloud data repositories. It provides several hundred integrations with SIEM (e.g., Splunk), SOAR (e.g., Palo Alto XSOAR), firewalls, VPNs, web proxies, DNS services, Active Directory, Entra ID, Microsoft Purview Information Protection, and Okta. The product includes a managed DDR service that uses behavioral detection models and automated remediation. Varonis has been on an acquisition binge to broaden its security platform to include better phishing, compliance testing, and AI protection. Varonis AWS pricing is $750 per TB per year.
Veeam DSPM
Since acquiring Securiti, Veeam has added a variety of breach and compliance management features to Veeam DSPM, as well as on-premises protection to complement its data backup tooling. Its tool supports data streaming technologies such as Confluent, Google PubSub, Kafka, and Kinesis. It comes with 350 content classifiers that support multiple languages along with more than a thousand predefined detection rules, including AI-based data sources and uses. It integrates with a wide collection of cloud-native security services, cloud access security brokers (CASBs), CNAPPs, CSPMs, cloud infrastructure entitlement management (CIEM) systems, DLP systems, intrusion detection systems (IDSes), Kubernetes security posture management (KSPM) systems, SIEM systems, and compliance tools. It is priced per terabyte, starting at $450 per TB per year for structured data and $1,000 per TB per year for unstructured data, with volume discounts available.
Wiz for DSPM
Wiz maintains a solid brand and product identity, despite being acquired by Google. It packages its products differently from most vendors, offering three products to protect code, cloud assets, and to defend against threats. All three are needed for a complete DSPM solution, which has been expanded to cover shadow data detection and AI-driven data classifiers. Wiz offers two licensing plans, but the full collection of DSPM features is available only on its more expensive Advanced plan. Wiz adds a lightweight agent called Runtime Sensor for detection and response. In addition to the usual cloud data sources, it also scans a variety of on-premises databases, such as MongoDB, MySQL, and PostgreSQL, as well as cloud versions, including Databricks. Wiz also integrates with more than 60 security products. Wiz’ AWS pricing starts at $38,000 per yr to protect 100 workloads.
View the full article
CSOonline

More All-Black Apple Vision Pro Parts Surface Online

Images of parts designed for an unreleased all-black Apple Vision headset have been leaked online, courtesy of X account @LusiRoy8.


The image shows what appear to be power strap and audio pod parts that look identical to Apple's existing Apple Vision Pro hardware, except with a dark finish that is not commercially available. The account that shared the images claims that the parts are for a "upcoming" second-generation Apple Vision Pro in black.

It's not the first time we've seen black versions of hardware related to Apple's headset, with images surfacing of similar parts last year. The leaker of those earlier images claimed that Apple has been testing a thinner and lighter mixed-reality headset referred to internally as "Vision Air," featuring a Midnight-colored exterior and reduced weight achieved by switching several structural components and the battery enclosure to titanium.

Apple was widely expected to launch both a lower-cost headset, tentatively dubbed "Vision Air," and a redesigned second-generation Vision Pro. However, Bloomberg's Mark Gurman reported in October that the company had paused development of all Vision headsets to focus on accelerating work on AI-powered smart glasses.

Gurman recently resisted reports that Apple has walked away from the headset entirely. The reporter says Apple hasn't fully abandoned the Vision Pro, but anyone hoping for a successor will be waiting at least two more years.

Indeed, Apple's smart glasses project is now the focus, and former Vision Products Group members have been reassigned to that team, as well as shoring up its Siri chatbot development. Apple is also busy working on other AI wearables such as the AirPods with cameras and a planned AI pendant.

Apple refreshed the Vision Pro in October 2025 with an updated model featuring an M5 chip.Related Roundup: Apple Vision ProBuyer's Guide: Vision Pro (Neutral)Related Forum: Apple Vision Pro
This article, "More All-Black Apple Vision Pro Parts Surface Online" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Microsoft previews automatic device isolation in Defender for Endpoint

Microsoft is previewing a new automatic device isolation capability in Defender for Endpoint’s auto attack disruption tool to help security pros contain cyber attacks in progress on their IT networks.
The company announced the capability earlier this month in a column about new features in Defender. There’s no word on when automatic device isolation will be in full production.
However, a new SANS Institute research paper warns that, in certain conditions, an attacker could leverage the new function to disable all user accounts.
The lesson, said Johannes Ullrich, the institute’s dean of research, is that autonomous AI action tools have to be tuned and tested like any other automation capability.
“Automatic isolation and attack disruption are not new concepts,” Ullrich said in an email, “but ideas like these have been used in the past in open source and commercial tools. This feature is most important in organizations with under-resourced IT security teams, as it automates attack response. However, these features must be carefully tuned. If they are left unconfigured, attackers can use them to delay response by disrupting accounts used by administrators.”
Nonetheless, in today’s environment, tools like these are important. Robert Enderle, IT consultant and head of the Enderle group, noted that modern automated malware and ransomware attacks move at machine speed, which means human response times are effectively obsolete.
By the time an analyst even sees a red flag, he said, the attacker has already established persistence or started encrypting files. Microsoft’s automatic device isolation acts as “a rapid, logical air gap. It instantly severs the device’s network connections, cutting off the attacker’s command and control (C2) and halting data exfiltration dead in its tracks. You have to bring an automated defense to an automated fight.”
He said a secondary benefit, often the more critical one for enterprise survival, is containing the blast radius. Attackers invariably use a compromised PC as a beachhead to move laterally across the corporate network, hunting for higher-value targets like domain controllers, he pointed out.
“By instantly quarantining that initial endpoint, you trap the threat where it stands. You ensure a single compromised laptop doesn’t metastasize into an enterprise-wide catastrophe,” he said.
There’s also is a massive forensic advantage, Enderle added. “In the old days, the instinct was often to literally pull the power plug, which destroys critical volatile memory, or physically yank the network cable, which completely blinds your remote security team. Logically isolating the device while maintaining a secure lifeline to security services preserves the crime scene. It prevents the attacker from deploying wiper malware or destroying logs, and it gives the Security Operations Center (SOC) the breathing room they need to safely investigate and remediate the machine without the panic of an actively spreading infection.”
How automatic attack disruption works
Automatic attack disruption is offered to organizations that subscribe to Microsoft Defender XDR, a unified cloud-based security suite that detects and investigates cyberattacks against PC, server, and IoT endpoints. It also manages hybrid identities and protects email and collaboration tools. As such, it correlates data to identify and respond to attacks.
The soon-to-be-delivered auto-isolation capability blocks most network traffic while keeping the device connected to security services. The action is time-limited and scoped to the incident, Microsoft said; security operators can release isolation at any time.
The broad automatic attack disruption capability uses AI to limit attackers’ lateral movement. “Attack disruption uses the full breadth of our extended detection and response (XDR) signals, taking the entire attack into account to act at the incident level,” Microsoft said in a detailed column describing the tool. “This capability is unlike known protection methods such as prevention and blocking based on a single indicator of compromise.”
To use automatic attack disruption, IT has to, at the least, enable Microsoft Defender for Endpoint Plan 2. It becomes more effective if Defender for Identity, Defender for Office 365 and Defender for Cloud apps are also deployed. Admins also have to configure appropriate permissions and monitoring.
Possible operational disruption
The SANS Institute’s academic paper by student Marcio Enriquez noted that AI systems that perform autonomous decisions like containment do improve response times and scalability. But they also rely on threshold-based logic derived from telemetry. “Even when operating on enterprise-wide data, they do not consistently account for system-level impact in their enforcement decisions,” the paper said, and thus can cause unintended disruptions when activated at scale. “This creates a gap between the need for rapid defensive actions and the organization’s ability to maintain operational continuity.”
It examined that gap by evaluating how threshold-driven autonomous containment actions can result in what it refers to as “large-scale operational disruption.”
Enriquez saw an example of this during a real security incident in the spring of 2025. A user in an organization was fooled by a phishing message and entered their credentials on a malicious website. Defender detected this, and within minutes initiated automated containment measures, including disabling the affected account, forcing a password reset and restricting logins across multiple managed devices.
However, because security analysts didn’t realize this was automated enforcement, they initially thought there had been lateral movement or widespread compromise. That triggered an emergency escalation involving security leadership, until further investigation realized that the propagation of containment controls was due to Defender.
“The event demonstrates the effectiveness of autonomous containment in rapidly interrupting active threats,” wrote Enriquez. “At the same time, it illustrates how automated response actions can generate enterprise-wide operational effects that are not immediately transparent to human operators.”
Could be weaponized
To test the ability of a threat actor to take advantage of a weakness in Defender XDR’s automatic attack disruption capability, Enriquez created a hybrid enterprise environment with 18 “users” and executed adversarial activity simulating hands-on-keyboard behavior across multiple identities to trigger high-confidence detection thresholds in Defender, through an attack tactic he calls Autonomous Defense Induced Disruption (ADID). In essence, it tricks the automatic disruption capability of Defender into giving a high-confidence score that the network is under attack.
“The results showed that when detection confidence thresholds were met, automated actions disabled all [18] Active Directory identities, including the local domain administrator, rendering the domain inaccessible,” Enriquez wrote.
“The research highlights the need for governance controls, privilege-aware safeguards, and system-level constraints to prevent autonomous containment from causing operational disruption,” he concluded.
Microsoft guidance: Keep auto attack disruption enabled
A Microsoft spokesperson said that the company has no comment on the research paper.
However, they said that Microsoft’s guidance is to keep automatic attack disruption enabled by default. “Opting out materially increases risk, particularly for multi-domain, multi-stage attacks such as HumOR [human intelligence operations, like social engineering], BEC [business email compromise] and AiTM [adversary in the middle], where even minutes of additional dwell time can translate into significant business impact.”
“At the same time,” Microsoft noted, “we recognize that security teams require control over autonomous actions. That’s why the capability is designed with granular controls. Security administrators can tune automation levels by device group and selectively exclude users, devices, or IP ranges based on operational needs. The recommended approach is targeted, intentional configuration, not a blanket opt-out. Customers retain full visibility into actions taken and have the ability to reverse automated responses at any time.”
View the full article
CSOonline

Apple Releases New Firmware for AirTag 2

Apple today released new firmware for its second-generation AirTag item trackers. The firmware has a 3.0.49 version number, up from 3.0.45, and it is the second firmware update that Apple has provided for the ‌AirTag‌ 2.


There is no word yet on what's included in the firmware, but ‌AirTag‌ firmware updates most often address bugs and make other under-the-hood improvements. Apple's prior firmware update tweaked the unwanted tracking sound to make it easier to find an unknown ‌AirTag‌ when using Precision Finding.

Apple will provide details on what's in the 3.0.49 firmware when it updates its firmware support page.

For the original ‌AirTag‌, firmware was distributed on a rolling basis over two weeks, but Apple appears to be pushing firmware updates to all ‌AirTag‌ 2 users at the same time.

You can check your AirTag firmware by opening the Find My app, going to the Items tab, selecting an ‌AirTag‌ in the list, and tapping on the ‌AirTag‌'s name to see its firmware version.

There is no way to force an ‌AirTag‌ update, and firmware is installed over the air via a connected iPhone. To get new firmware, make sure your ‌AirTag‌ is within range of your iPhone, and then wait for the firmware to roll out.Related Roundup: AirTagBuyer's Guide: AirTag (Buy Now)
This article, "Apple Releases New Firmware for AirTag 2" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

iOS 27 Siri Redesign Will Use Dark Color Scheme, Matching Apple's WWDC Art

Apple is redesigning Siri for iOS 27 to accommodate new artificial intelligence features and chatbot capabilities. ‌Siri‌ is getting a dedicated app, integration with the iPhone's Dynamic Island, and a new design scheme.


The graphics that Apple is using to promote WWDC hint at its design plans for ‌Siri‌, according to Bloomberg's Mark Gurman. In this week's Power On newsletter, Gurman said the logo Apple uses for its WWDC website features the same colors it will use for ‌Siri‌ animations and input fields in ‌iOS 27‌.

Apple's WWDC site features the Swift bird logo in white on a black background, with subtle highlights in pink, dark blue, purple, and orange. The colors are reminiscent of the current ‌Siri‌ animation that surrounds the iPhone's display when ‌Siri‌ is activated, but the shades are softer and not as saturated.

The ‌Siri‌ interface that Apple is testing uses all dark colors with no light mode available for now. Several of the new ‌Siri‌ UI elements will have a dark background with a cursor that blinks in those same colors.

Apple is creating a dedicated app for ‌Siri‌ for ongoing chatbot-style conversations, and it'll look similar to existing chatbot apps but with a Messages-style aesthetic. When ‌Siri‌ is activated, there will be a pill-shaped animation in the ‌Dynamic Island‌ and a glowing "searching" label while ‌Siri‌ is answering a query. Results are displayed in a translucent panel, and pulling down on the panel will initiate a conversation interface.

Swiping down from the top center of the display will activate a system-wide search interface with a Search or Ask bar for typing or speaking questions to ‌Siri‌.

Apple plans to entirely overhaul ‌Siri‌, and the personal assistant will be able to do far more than before. Apple has licensed Google's Gemini models to power ‌Siri‌ after its own AI models proved inadequate. With Gemini as ‌Siri‌'s backbone, Apple should be able to match many of the AI features that Google offers.

‌iOS 27‌ will be introduced at the WWDC 2026 keynote on June 8.Related Roundups: iOS 27, WWDC 2026Tag: SiriRelated Forum: Apple, Inc and Tech Industry
This article, "iOS 27 Siri Redesign Will Use Dark Color Scheme, Matching Apple's WWDC Art" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Level Lock Pro Review: An Apple Home Key Smart Lock That Doesn't Look Like One

The Level Lock Pro is Level's latest smart lock, featuring Matter connectivity for Apple Home, multiple unlocking methods, door status, and the unassuming design that Level products are known for.


I've tested and reviewed several great HomeKit-compatible smart locks, but Level Locks are my personal favorite because of the look. From both the inside and the outside, Level Locks look like a standard deadbolt and not like a smart lock. I had feature-rich smart locks from Aqara that I was using for about a year after a review, but I got tired of looking at the bulky boxes on my doors. A couple of months ago, I bought two standard Level Locks, and then later, Level sent me the Level Lock Pro.

I don't think there's any smart lock solution that has a better aesthetic than the Level Lock, so if that's important to you, these are the locks to get. It comes in satin nickel and matte black, so it should match many standard doorknobs. The Level Lock Pro has an IP54 water and dust resistance rating, so it will hold up fine in the rain.


Level Locks are not the cheapest locks on the market, and depending on what you're comparing against, there's a premium for design. The Level Lock Pro is $349, and the Level Lock is $249. Aqara locks range from $150 to $270, and Matter locks from Eufy, Yale, and Kwikset are in that same range.

The Level Lock Pro replaces a standard deadbolt and strike plate on your door, so installation is a matter of pulling out the existing deadbolt and walking through the Level Lock Pro instructions to install the new lock. I am going to blame this on my crummy doors, but I have more trouble installing Level Locks than other smart locks. Level Locks have a wide, circular bolt that's not the shape of most deadbolts, and I haven't had a Level Lock setup where I didn't have to fuss with the fit of the lock in the door or the fit of the plate on the doorframe. I generally get things to work, but there's frustration involved.


There are smart locks that can unlock your door with fingerprint sensors, palm recognition, facial scans, and codes, but the Level Lock Pro is simpler. You can use a key, one of the two included NFC key fobs, tap to unlock with your phone or watch, use the Home app or Level app, or ask Siri to unlock the door.

The Level Lock Pro integrates with HomeKit using Matter, and it also supports Apple Home Key so you can store a key in the Wallet app on iPhone or Apple Watch. With Home Key, I can unlock my door without having to unlock my iPhone and with no need for Face ID. I just tap my phone or my watch on the lock, and it unlocks.


‌Siri‌ and the Home app work for unlocking too, and there's a Level app. I don't use the Level app, but it is available for locking and unlocking, assigning codes, setting up auto lock and auto unlock (which uses Bluetooth and unlocks when you're in range), adjusting sound, giving someone a door code, and enabling door status. Like the Level Lock, the Level app has an uncomplicated design, so it's easy to get to all of the features.

Door status is a Level Lock Pro feature that lets you know if your door is open or closed, and it works when the door is unlocked. I have the Level Lock Pro on my garage door, and it's a door that's often not locked, so it's useful to get an alert when it's opened.

I use the Home app and ‌Siri‌ to unlock my Level Locks, especially if I'm not home and need to let someone in. I also ask ‌Siri‌ to open the door as I approach, so a lot of the time, I'm not even using tap to unlock. The Home app sends a notification to my iPhone and Apple TV when a connected lock is locked or unlocked, and the Home app Activity log keeps track of when each door was locked or unlocked. Everyone that's invited to an Apple Home can access the lock, but you can also share access with the Level app. The Level app supports temporary entry, which is useful for a one-time event or a weekly cleaning.


The Home app is also useful for automations, like locking up automatically when everyone leaves the home or unlocking the door at a certain time. I have an automation that locks all my locks at 10:00 p.m., just in case I forget to lock one of the doors.


For remote access features, you need a Matter-over-Thread controller and a border router, which are requirements fulfilled by a HomePod or ‌Apple TV‌. You need one of those to add any Matter-enabled device to ‌HomeKit‌. The Level Lock Pro connects to Apple Home using Thread instead of Wi-Fi, but if you want Wi-Fi connectivity, there is an optional Level Connect Wi-Fi Bridge. I haven't needed it because ‌HomeKit‌ provides all of the same functionality. You can also add on a keypad if you want that option.


Most smart locks have a battery in the box that goes on the door, but the Level Lock Pro's battery is in the deadbolt. It uses a CR2 Lithium battery, which fits inside the deadbolt once the cap is taken off. Changing the battery is a matter of opening the door, locking it, popping out the old battery, and adding in the new one. The Level app lets you know battery status, so you can keep tabs on when it's time to update the battery. Each battery lasts for about a year, and I haven't had to change mine yet.

According to Level, the Level Lock Pro has an ANSI Grade 1 bump- and pick resistant cylinder, which isn't common for smart locks. That means it's resistant to lockpicking, it's harder to drill out, and lock bumping is harder.

How to Buy

The Level Lock Pro is available from the Level website or from Amazon.com for $349.

Note: Level provided MacRumors with Level Lock Pro for the purpose of this review. No other compensation was received.
Tag: Apple Home
This article, "Level Lock Pro Review: An Apple Home Key Smart Lock That Doesn't Look Like One" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple's Beautiful Barcelona Store Reopens With Pickup Station and More

Apple's beautiful Passeig de Gràcia store in the heart of Barcelona reopened today, after being closed for around three months for renovations.


According to the Spanish blog Applesfera, the store's large video wall has been replaced with a dedicated Apple Pickup station for online orders. The indoor trees and wood cube seats that surrounded the screen have also been removed. With these fixtures removed, the store's iconic glass staircase is more visible again.

In addition, the store's terrazzo floor has received a brighter white finish.

Apple Passeig de Gràcia's main floor after remodeling (via Applesfera)
Apple Passeig de Gràcia first opened in 2012, and it is one of the company's flagship retail locations. The store is on one of the most popular avenues in Barcelona, inside a historic former bank building with a stunning stone facade.Tag: Apple Store
This article, "Apple's Beautiful Barcelona Store Reopens With Pickup Station and More" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Earn a Running Day Apple Watch Activity Award on June 3

Apple plans to hold an Apple Watch Activity Challenge to celebrate Global Running Day on Wednesday, June 3.


To complete the challenge, Apple Watch owners will be required to record a running workout of at least 5K on Global Running Day.

As a reward, Apple Watch owners can unlock a dedicated award in the Fitness app, plus animated stickers that can be used in the Messages app.








Apple has been celebrating Global Running Day since 2024, and it comes after the April 2026 Earth Day and International Dance Day Activity Challenges.Related Roundup: Apple Watch 11Tag: Activity ChallengeBuyer's Guide: Apple Watch (Caution)
This article, "Earn a Running Day Apple Watch Activity Award on June 3" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

iOS 26.6 Will Alert You When You've Maxed Out Blocked Contacts

Apple's iOS 26.6 update appears to add new wording around blocked contact limits, though it is unclear if the actual limits have changed.


Code in the beta suggests users will get a warning if they exceed the maximum number of blocked contacts. "You've reached the maximum number of blocked contacts. To block additional callers, remove a blocked contact in Settings," reads the alert, which is titled "Blocked Contacts Limit Reached."

Based on discussions on social media and Apple's Support Communities, some users have been unable to block additional contacts after hitting a 20,000 limit. Other people have mentioned running into issues after 8,000, and some have experienced issues with even fewer phone numbers blocked. Apple does not offer documentation on blocking limits.

With limits in the thousands at least, it's unlikely most people have had blocking problems, though a person who is blocking spam callers regularly could eventually hit a cap. iOS 26.6 might make it clearer when a limit has been reached, and what to do about it.

Removing older blocked contacts is the solution, which can be done by going to Settings > Apps > Phone > Blocked Contacts. There is no bulk unblocking tool, and the easiest way to remove a contact is to swipe left on each entry. Alternatively, you can select Edit, tap on the red minus button next to each contact, and choose the unblock option.

iOS 26 added an Ask Reason for Calling option that sends calls from people who aren't in your Contacts directly to voicemail, which is an easier option for spam call management than blocking phone numbers. With the feature turned on, a caller can state their reason for calling and the person receiving the call can decide whether to pick up. Alternatively, all calls from unknown numbers can be silenced and sent to voicemail with no alert using the Silence option.

Missed calls and voicemails from unknown callers can also be filtered into a separate Unknown Callers list in the Phone app. Some carriers also offer a separate spam detection option that can send calls from known spammers to the Spam list.

Apple seeded the first beta of iOS 26.6 to developers today, and the software may soon be made available to public beta testers. A public release is likely several weeks away. So far, there are no other known features in iOS 26.6.Related Roundups: iOS 26, iPadOS 26Related Forum: iOS 26
This article, "iOS 26.6 Will Alert You When You've Maxed Out Blocked Contacts" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

First macOS Tahoe 26.6 Beta Now Available for Developers

Apple today provided the first beta of an upcoming macOS Tahoe 26.6 update to developers for testing purposes, with the update coming two weeks after Apple launched ‌macOS Tahoe‌ 26.5.


Developers can download the ‌macOS Tahoe‌ 26.6 update by opening up the System Settings app, selecting the General category, and then choosing Software Update. Beta Updates will need to be enabled, and a free developer account is required.

With macOS 27 set to be unveiled in less than a month, Apple is likely focusing most of its attention on the new software. We are not expecting any major new features in ‌macOS Tahoe‌ 26.6.

The beta is limited to developers right now, but a public beta is expected in the next week or two.Related Roundup: macOS TahoeRelated Forum: macOS Tahoe
This article, "First macOS Tahoe 26.6 Beta Now Available for Developers" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Seeds First iOS 26.6 and iPadOS 26.6 Betas to Developers

Apple today seeded the first betas of upcoming iOS 26.6 and iPadOS 26.6 updates to developers for testing purposes, with the software coming two weeks after Apple released iOS 26.5 and iPadOS 26.5.


Registered developers can download the betas from the Settings app on the iPhone or iPad by going to the General section and selecting Software Update.

With the debut of iOS 27 approaching in early June, Apple is wrapping up work on iOS 26. We are not expecting any major new features in the iOS 26.6 update, and it will likely focus on bug fixes and performance improvements. Related Roundups: iOS 26, iPadOS 26Related Forum: iOS 26
This article, "Apple Seeds First iOS 26.6 and iPadOS 26.6 Betas to Developers" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Releases First watchOS 26.6, tvOS 26.6 and visionOS 26.6 Betas

Apple today provided developers with the first betas of upcoming watchOS 26.6, tvOS 26.6, and visionOS 26.6 betas for testing purposes. The software two weeks after Apple launched the 26.5 versions of each platform.


The software updates are available through the Settings app on each device, and because these are developer betas, a free developer account is required.

There's no word on what's in the software as of yet. watchOS, tvOS, and visionOS often get few features in each new beta, with updates primarily focusing on bug fixes and performance improvements.

Apple will likely provide public beta testers with access to the tvOS 26.6 and watchOS 26.6 betas in a week or two, but visionOS 26.6 will remain limited to developers.Related Roundups: Apple TV, Apple Vision Pro, watchOS 26Buyer's Guide: Apple TV (Don't Buy), Vision Pro (Neutral)Related Forums: Apple TV and Home Theater, Apple Vision Pro, Apple Watch
This article, "Apple Releases First watchOS 26.6, tvOS 26.6 and visionOS 26.6 Betas" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Android Brands May Copy Apple's New Split iPhone Launch Strategy

Android manufacturers are planning to adopt Apple's split launch strategy, releasing high-end and standard models in separate windows rather than simultaneously, according to the leaker known as "Digital Chat Station."


The leaker made the claim in a new post on Weibo this week, saying the "Android camp may repeat this style of play" with Pro series and standard models launching separately in a move to "comprehensively go head-to-head" with Apple. The leaker described it as a move to "fully benchmark" Apple, suggesting the motivation is competitive rather than logistical.

The same post reiterated earlier predictions about Apple's plans. Starting this year, Apple is widely expected to break from its long-standing September release cycle by splitting the iPhone 18 lineup across two windows: the iPhone 18 Pro, ‌iPhone 18 Pro‌ Max, and the first foldable iPhone are expected to launch in fall 2026, while the standard ‌iPhone 18‌, iPhone 18e, and a second-generation iPhone Air are expected in spring 2027.

Digital Chat Station attributed the delay partly to supply pressure on memory and 2nm chip production, which is an explanation consistent with Nikkei Asia's corroborating report in January, which also cited a deliberate commercial motive to maximize revenue from premium models before cheaper alternatives arrive.

Supply chain analyst Ming-Chi Kuo and The Information have also supported the rumor of the split launch. Kuo framed the strategy as a way to prevent "diluted marketing efforts" as Apple's lineup expands to six devices and to address the "marketing gap" created by Chinese Android brands that typically launch their flagships in the first half of the year, a window Apple has historically ceded entirely to Android.

If Android brands do adopt the same release plan, it would mark a noticeable departure from current practice. Samsung, Apple's most direct competitor, launches its Galaxy S flagship family, standard, Plus, and Ultra, simultaneously each February or March, then launches foldables in a separate mid-year event in July. All tiers of the S series ship together and there is no equivalent of deliberately holding the base model back.

Xiaomi regularly launches flagship models in China several months before a global rollout, and its Ultra-tier models often arrive weeks or months after standard and Pro variants within the same generation. Oppo and Vivo similarly stagger Ultra devices relative to their base flagships, but in each case the split is led by entry level models debuting first, followed by high-end devices.

Should Android manufacturers adopt Apple's new plan, it would largely represent an inversion of the current approach, with premium models leading and standard devices following months later.Tags: Android, Digital Chat Station
This article, "Android Brands May Copy Apple's New Split iPhone Launch Strategy" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple Watch for Diabetes: The Latest on Apple's Plans for Non-Invasive Blood Sugar Monitoring

For many years now, it has been rumored that the Apple Watch will eventually gain non-invasive blood sugar monitoring capabilities, which would enable millions of people with diabetes to track their blood glucose levels without needing to prick their skin with a needle or wear a dedicated continuous glucose monitor.


According to Bloomberg's Mark Gurman, Apple recently shifted oversight of the project from its platform architecture chief Tim Millet to Zongjian Chen, a senior engineer overseeing advanced technologies within the company.

He framed this change as positive news for the project, which has apparently been in development for more than 15 years.

"Some view the transition as a sign the work may finally be progressing to a point where Chen, known as someone who delivers, can ramp up development of the technology into an eventual consumer-grade offering," he said.

In 2023, Gurman reported that Apple's system would rely on a laser that would emit light under the skin to determine a person's blood glucose level.

"The system uses lasers to emit specific wavelengths of light into an area below the skin where there is interstitial fluid — substances that leak out of capillaries — that can be absorbed by glucose," he said. "The light is then reflected back to the sensor in a way that indicates the concentration of glucose."

An algorithm would ultimately determine a person's blood glucose level, and the feature could also alert users to potential signs of prediabetes.

While the project has new leadership, the Apple Watch is still unlikely to gain non-invasive blood sugar monitoring for several more years, if ever. But if Apple eventually achieves this moonshot, the Apple Watch would provide diabetic people with a more comfortable and convenient solution for keeping track of their blood sugar.Related Roundup: Apple Watch 11Tags: Bloomberg, Mark GurmanBuyer's Guide: Apple Watch (Caution)
This article, "Apple Watch for Diabetes: The Latest on Apple's Plans for Non-Invasive Blood Sugar Monitoring" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

MuddyWater Uses DLL Side-Loading in Espionage Campaign Targeting 9 Countries

The Iranian hacking group known as MuddyWater has been linked to a new campaign affecting at least nine organizations across nine countries on four continents in the first quarter of 2026. The activity targeted industrial and electronics manufacturing, education and public-sector bodies, financial services, and professional services, per the Threat Hunter Team from Symantec and Carbon Black.View the full article
Hacker News

Leaker Pushes Back on Rumors of Pro iPhone's Return to Titanium

The Pro iPhone models are unlikely to return to titanium in the near future due to the heat dissipation demands of local AI, according to a known Weibo leaker.


The claim comes from the leaker known as "Fixed Focus Digital," and pushes back on an earlier report from "Instant Digital," who suggested Apple was weighing up the use of liquid metal or an improved titanium alloy as a longer-term replacement for aluminum iPhone frames. Fixed Focus Digital argues that aluminum's thermal properties make it the only practical choice for now, given the processing requirements of AI features. The leaker adds that this is not an Apple-specific issue, noting that Android and Huawei HarmonyOS devices also prioritize aluminum for the same reason.

Instant Digital's earlier report argued that Apple's switch from titanium to aluminum for the iPhone 17 Pro was a compromise solution while it continued to develop longer-term alternatives. The leaker claimed Apple was exploring both liquid metal and revised titanium alloys for future Pro models, with both materials reportedly already earmarked for the upcoming foldable iPhone.

Apple switched away from titanium following overheating complaints on the iPhone 15 Pro and iPhone 16 Pro models, although the iPhone Air continues to use it. Fixed Focus Digital's assessment suggests aluminum is more deeply entrenched in Apple's plans than Instant Digital's framing implied, at least for the foreseeable future. The iPhone 18 Pro is expected to retain the same aluminum unibody design as the iPhone 17 Pro models, meaning any material change is unlikely before 2027 at the earliest.Tag: Fixed Focus Digital
This article, "Leaker Pushes Back on Rumors of Pro iPhone's Return to Titanium" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

GitHub Actions abused by Megalodon attack to slip malicious commits into 5,500 repos

A large-scale automated GitHub backdooring campaign was caught pushing thousands of malicious commits into public repositories while posing as routine CI/CD upkeep.
Researchers at SafeDep observed the campaign, Megalodon, touching more than five thousand repositories over a six-hour window on May 18. The attack was in the form of a malicious commit, “acac5a9,” targeting GitHub Actions workflows.
Unexpected workflow_dispatch runs in the Actions tab could be a warning sign, the researchers said in a blog post. “If you use OIDC federation for cloud deployments, review cloud audit logs for token requests from unknown workflow runs.”
The malicious commits were seen modifying Github Actions workflows to include base64-encoded bash payloads designed to steal secrets exposed during CI execution, including cloud credentials, SSH keys, OpenID Connect (OIDC) tokens, source code secrets, and other environment variables.
Among the hardest-hit projects were Wiznet’s ioLibrary_Driver repository, four Tiledesk repositories, and four persian-tools repositories, with well over 2,000 malicious commits between them.
A later blog post by OX Security flagged some similarities to the widespread TeamPCP compromises, particularly the use of hardcoded historical commit dates. This was a trick used in TeamPCP-linked operations to hide the true timing of malicious activity.
An automated attack with compromised keys
SafeDep’s researchers said the Megalodon campaign pushed 5,718 malicious commits across 5,561 public GitHub repositories within roughly six hours, primarily by abusing compromised credentials to directly modify GitHub Actions workflows.
It detected the campaign while investigating a Tiledesk GitHub Actions workflow file hiding the base64-encoded bash payload.
“Versions 2.18.6 (May 19) through 2.18.12 (May 21) all carry the backdoor,” the researchers had concluded after preliminary investigation.
Further comparison of the culprit Tiledesk versions with their legitimate predecessors led the researchers to a malicious commit . “The malicious commit landed on May 18, 2026, authored by build-bot <[email protected]> with the message ‘ci: add build optimization step’,” they said. “The author name and generic noreply email mimic automated CI commits.”
The commit was pushed without a pull request (PR) or a merge commit, which the researchers said was done using a compromised Personal Access Token (PAT) or deploy key. They warned the malicious commit remained active on the “master branch” at the time of publishing the disclosure.
Besides “build-bot”, the campaign also relied on other forged author identities such as “auto-ci” and “ci-bot.”
The campaign pushed out two payload variants. The first, “SysDiag,” embedded obfuscated bash payloads directly into workflows, triggering automatically on every push or PR, while the second, “Optimize-Build,” followed a staged approach using ‘workflow_dispatch” to execute the malicious workflow only when needed. The latter, used in the Tiledesk compromise, proved operationally noisy and left detectable traces.
Both variants targeted sensitive CI secrets including AWS and GCP credentials, SSH keys, Kubernetes configs, GitHub OIDC tokens, source code secrets, and shell history. They exfiltrated stolen secrets to attacker-controlled infrastructure at 216.126.225.129:8443.
SafeDep shared a list of indicators of compromise (IOCs) including the C2 domain, campaign signature, author names and emails, commit messages, and the names of the compromised GitHub repositories to aid in detection and clean-up.
View the full article
CSOonline

iPhone 18 Pro Launching Later This Year With These 10 New Features

While the iPhone 18 Pro and iPhone 18 Pro Max are not launching until September, there are already plenty of rumors about the devices.


It was initially reported that the iPhone 18 Pro models would have fully under-screen Face ID, with only a front camera visible in the top-left corner of the screen. However, the latest rumors indicate that only one Face ID component will be moved under the screen on the devices, which will result in merely a smaller Dynamic Island.

Below, we have recapped 10 features rumored for the iPhone 18 Pro models, as of May 2026:Dark Cherry: The special color for the iPhone 18 Pro models will reportedly be Dark Cherry, alongside Light Blue, Dark Gray, and Silver. The existing Cosmic Orange and Deep Blue colors are expected to be discontinued.
Smaller Dynamic Island: It has been rumored that Face ID's flood illuminator will be moved under the screen on the iPhone 18 Pro models, paving the way for a smaller Dynamic Island on the devices.
LTPO+ Displays: The next Pro models are expected to have the same overall design as the iPhone 17 Pro models, including 6.3-inch and 6.9-inch display sizes and a "plateau" housing three rear cameras. However, the displays will reportedly use so-called LTPO+ display technology, which should contribute to longer battery life.
Variable Aperture: The main 48-megapixel Fusion camera on both iPhone 18 Pro models is rumored to have a variable aperture, which would allow users to control the amount of light that passes through the camera's lens and reaches the sensor. This would provide greater control over depth of field. However, given that iPhones have smaller image sensors due to smartphone size constraints, it is unclear exactly how meaningful this improvement would be.
A20 Pro Chip: Apple's next-generation A20 Pro chip is expected to use TSMC's first-generation 2nm process, whereas the A19 Pro chip is 3nm. With a 2nm architecture and a new packaging design, the A20 Pro chip should deliver solid year-over-year performance and power efficiency gains.
C2 Modem: Apple's custom C1 cellular modem for 5G and LTE debuted in the iPhone 16e last year, and that was followed by a C1X chip in the iPhone Air. Apple says the C1X modem is up to twice as fast as the C1 modem, and the most power-efficient modem in an iPhone ever. The improvements should continue with Apple's third-generation C2 modem in the iPhone 18 Pro models.
5G via Satellite: With the C2 modem, the iPhone 18 Pro models will reportedly support 5G via satellite for web browsing without Wi-Fi or cellular connectivity.
N2 Chip: Most of the iPhone 17 models and the iPhone Air are equipped with an Apple-designed N1 chip that enables Wi-Fi 7, Bluetooth 6, and Thread. Apple says the N1 chip also improves the overall performance and reliability of features like Personal Hotspot and AirDrop. iPhone 18 Pro models are expected to have Apple's next-generation N2 chip, but it is not yet known what improvements would come with this upgrade.
Simplified Camera Control: Apple is expected to simplify the Camera Control button on the iPhone 18 Pro models, by removing touch sensitivity and haptic feedback. The redesigned button will only have pressure sensitivity.
Redesigned Rear Ceramic Shield: The rear Ceramic Shield area for MagSafe is rumored to feature a more frosted and seamless appearance on the iPhone 18 Pro models compared to the current two-tone design.Apple is expected to release the iPhone 18 Pro, iPhone 18 Pro Max, and a foldable iPhone in September, followed by a standard iPhone 18 model, a lower-end iPhone 18e, and a second-generation iPhone Air early next year.Related Roundup: iPhone 18 Pro
This article, "iPhone 18 Pro Launching Later This Year With These 10 New Features" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Apple's M5 MacBook Air Hits New Low Price of $899.99

Amazon today has introduced a new record low price on the 512GB 13-inch M5 MacBook Air, available for $899.99, down from $1,099.00. This deal is available in all colors and as of writing only Amazon has the discount.

Note: MacRumors is an affiliate partner with Amazon. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.

This new price is about $50 cheaper when compared to past sales. Amazon also has $149 off the 1TB models of the 13-inch M5 MacBook Air, which match all-time low prices on these devices. Delivery dates vary depending on the color selected, but most will arrive before the end of the month.

$199 OFF13-inch M5 MacBook Air (512GB) for $899.99

If you're on the hunt for more discounts, be sure to visit our Apple Deals roundup where we recap the best Apple-related bargains of the past week.



Deals Newsletter

Interested in hearing more about the best deals you can find in 2026? Sign up for our Deals Newsletter and we'll keep you updated so you don't miss the biggest deals of the season!




Related Roundup: Apple Deals
This article, "Apple's M5 MacBook Air Hits New Low Price of $899.99" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

The Untrusted Autonomous Workload: How AI Coding Agents Reshape What Isolation Has to Do

Earlier this year I mass-migrated my blog to Astro using Claude Code. 146 posts. 6,024 images. Canonical URLs, JSON-LD markup, sitemap generation, the whole stack. I’d spent hours writing a skills file to teach the agent about my blog’s architecture, how deployment worked, what not to touch. And it worked. Claude Code rewrote components, fixed trailing-slash mismatches across hundreds of pages, added BreadcrumbList structured data to hundreds of routes. Lighthouse scores hit 97 on performance. The blog looked better than it ever had.
The problem was that I had stopped understanding my own codebase.
Not completely. I could still read the files. But somewhere around the third round of “fix the error that the last fix introduced,” I caught myself copy-pasting stack traces back into Claude and trusting whatever came back. The agent would make a change, something else would break, I’d ask the agent to fix that too, and a few cycles later the blog worked again. I couldn’t have told you what was actually in the PostCSS config or why the GA4 integration was wired up the way it was. It worked. It looked great. My confidence in what was underneath had quietly evaporated.
That feeling (it works, thank god, let’s not touch it) is the feeling of having given an autonomous agent real access to your codebase. Every developer using these tools knows it. Nobody writes about it in vendor blog posts. And it’s what made me understand, on a level deeper than reading documentation, why Docker had to build Sandboxes.
Because here’s what I hadn’t thought about: while Claude Code was rewriting my Astro components and fixing image CLS across hundreds of files, every npm install it ran happened on my laptop. Same for every file it modified and every package it pulled. My user privileges, no boundary in sight. If the agent had decided to modify a Git hook or rewrite a CI workflow, I would not have noticed. I wasn’t reviewing individual file changes at that point. I was reviewing outcomes. And reviewing outcomes while skipping changes is not a security model. It’s a prayer.
Docker Sandboxes exists to close that gap.
The container model and why it doesn’t stretch here
Containers were never the wrong abstraction. They were the right abstraction for a world where you knew what was inside them. For twelve years that world held: you wrote the code, you reviewed it, you put it in a Dockerfile, and the container gave it a clean room to run in. Shared kernel was fine because the threat model was bugs in your own software, not surprises from a tenant you’d just invited in.
AI coding agents don’t fit. They aren’t bugs in your software because they aren’t your software. They’re a new kind of tenant, one that’s autonomous and privileged in ways that would make any security engineer nervous. The agent installs packages you didn’t pick and runs commands you didn’t script. It makes network calls you’d never have predicted, to endpoints you didn’t know were in your dependency tree. The trust profile is code being written right now, by something that won’t pause to ask permission. Containers were built for a different kind of code.
This isn’t hypothetical. On March 19, 2026, attackers force-pushed 76 of the 77 version tags in aquasecurity/trivy-action and published a malicious Trivy v0.69.4 binary to GitHub Releases. The exposure window was about 12 hours. The compromised code scraped CI runner memory for secrets, cloud credentials, SSH keys, and Kubernetes tokens, exfiltrating them to a typosquatted domain. Every pipeline that referenced trivy-action by version tag during that window ran code nobody on the receiving end had reviewed.
What gets me about Trivy: the weaponized tool was a vulnerability scanner. The thing organizations deployed to find malicious code became the malicious code. The maintainers didn’t write the bad binary; a compromised CI workflow with too much access and not enough containment did. Substitute “compromised CI workflow” with “AI agent in permissive mode” and you have the same threat model, running all day on every developer machine.
Containers were the right answer to “I trust this code, I want to run it cleanly.” They were never going to be the right answer to “I don’t fully trust this code, and I want to give it real work to do anyway.” That’s the gap microVMs fill.
What Docker built, and why each piece is there
First choice: don’t patch containers. There’s a long tradition in our industry of making a familiar abstraction handle a new problem by adding flags to it. Privileged mode, capability dropping, seccomp profiles, gVisor in front of runc. All of those have their place. None of them solved the specific issue that an autonomous agent needs its own Docker daemon. Docker-in-Docker either compromises the isolation (privileged mode, host socket mounting) or creates a nested complexity that becomes its own attack surface. The Docker docs are blunt about this. Containers, they say, share the host kernel and “can’t safely isolate something that needs its own Docker daemon.”
Once you accept that, you end up at a VM. Not a heavyweight one (booting Ubuntu Server for every coding session would be absurd) but a microVM: light enough to start in seconds, with just enough kernel to run the agent’s containers.
Docker Sandboxes uses a custom VMM, not Firecracker. If you’ve read the Firecracker spec and you’re thinking “boots in 125ms with under 5MB of overhead,” those are Firecracker’s numbers, not Docker’s. Different microVM implementations have different cost profiles. Platform specifics: Hypervisor.framework on macOS, Windows Hypervisor Platform on Windows, KVM on Linux.
Caption: The Sandbox architecture. Each microVM runs its own kernel and its own Docker Engine. Credentials never cross the VM boundary.
Inside each microVM, the sandbox runs a complete Docker Engine. When the agent runs docker build, that command goes to a private daemon that doesn’t know your host containers exist. When it pulls an image, the image lives inside the sandbox VM. When you delete the sandbox, the entire image cache goes with it. Multiple sandboxes don’t share layers. Wasteful. Worth it.
The first time I looked inside a running sandbox, the agent was running as root with sudo and full Docker Engine access inside the VM. My reflex was that this had to be wrong. You don’t give root to untrusted code. But the design is right: the isolation model doesn’t constrain what the agent does inside the boundary. It constrains where the consequences land. Inside the VM, the agent can do whatever it wants. Outside? Nothing. Trying to lock the agent down with capability dropping inside the VM would be solving the wrong problem. The agent legitimately needs to install packages and run docker build. What it doesn’t need is for any of that to touch your laptop.
Caption: From the host, sandboxes don’t show up in docker ps because they aren’t containers; sbx ls is how you see them.
The network layer is where it gets interesting, because it doubles as the credential boundary.
Outbound HTTP/HTTPS traffic routes through a proxy on the host, accessible from inside the VM at host.docker.internal:3128. UDP and ICMP are blocked at the network layer and can’t be allowed by policy. Non-HTTP TCP (like SSH) needs explicit IP+port rules. DNS resolution goes through the proxy. If a request can’t go through the proxy, it doesn’t leave. The proxy terminates TLS, inspects the host header, applies your policy, and re-encrypts with its own certificate authority that the sandbox trusts. Man-in-the-middle by design. Docker uses that exact framing in the documentation.
MITM is what makes credential injection work. Agents need API keys: for the AI provider, for registries, sometimes for cloud accounts. Naive answer is to pass those credentials in as environment variables, where they sit inside the VM and follow it everywhere. Docker instead keeps credentials on the host, in your OS keychain, and has the proxy inject them into outbound requests transparently. The agent sees requests that just work, and the VM never had the secrets to begin with. The docs don’t hedge on this: credential values are never stored inside the VM. A compromised sandbox can’t exfiltrate your API keys because your API keys were never in there.
Docker tells you what won’t work
Sandboxes documentation has a quality that’s rare in security architecture docs: it tells you what the system doesn’t protect against. Most of these documents are written to make a product look strong. Docker’s docs surface the limits. Two of them matter.
The first one is about the network policy.
At first sbx login, you pick one of three default policies. Open allows everything except blocked CIDR ranges (private networks, link-local addresses, cloud metadata endpoints). Balanced denies by default but pre-allows common dev domains. Locked Down denies everything until you explicitly allow. Locked Down is the strictest option, the deny-by-default mode you’d want if you were paranoid. But even with Locked Down and a curated allowlist, the proxy filters by domain, not by content.
Here’s the exact language from the docs: allowing broad domains like github.com permits access to any content on that domain, “and agents could use these as channels for data exfiltration.” Security vendors don’t usually say this about their own products. If github.com is on your allowlist (and it almost certainly is, because the agent needs to clone repos), the proxy knows the request is going to github.com. It does not know whether the agent is reading documentation, cloning a repository, or creating a public gist with the contents of your .env file. All three look identical at the domain level. Same goes for every allowlist entry that includes user-generated content: Discord webhooks, Notion pages. “The domain is allowed” doesn’t mean “only safe content lives there.”
Caption: Under a deny policy, non-allowlisted domains are blocked. Allowlisted domains succeed, including domains that host arbitrary user-generated content.
Docs also acknowledge domain fronting as an inherent limitation of HTTPS proxying. Proxy sees which domain a request claims to be going to; it cannot always prevent the request from being routed elsewhere through that allowed CDN.
The microVM boundary is the primary isolation. Network proxy is a useful additional control, especially for blocking accidental access to internal networks. It is not a hermetic seal, and Docker doesn’t claim it is. “The agent is on a deny policy” is not the same thing as “the agent cannot send data anywhere.”
The workspace is always shared
Network policy is the smaller honest limit. Workspace sharing is the bigger one.
The microVM boundary is strong everywhere except for one path that crosses it on purpose: the workspace directory.
The whole point of running an agent in a Sandbox is for the agent to do real work in your real codebase. Docker shares the workspace between the host and the sandbox at the same absolute path. When the agent edits a file inside the sandbox, the file changes on your host. When you pull a new commit on your host, the agent sees it. This is the design. It’s exactly what you want from a developer tool.
It’s also a covert channel that the agent has legitimate write access to.
Docker security documentation spells out what “the same files” includes, and this is what matters: files that execute implicitly during normal development. Git hooks. CI configurations. IDE task definitions. Makefile targets. package.json scripts. Pre-commit configs. Anything that runs when you do something that feels like just “using your tools.”
Simplest version of the attack: an agent inside the sandbox writes a malicious post-commit hook to .git/hooks/post-commit. Git hooks don’t appear in git diff. They live in .git/, which most developers never open. Next time you commit on your host, the hook runs on your host with your user privileges. Sandbox boundary doesn’t matter, because the boundary ended at the workspace, and the workspace was always shared.
Which brought me back to my own Astro migration, uncomfortably. I’d let Claude Code rewrite hundreds of files across my blog. I’d reviewed the outcomes (Lighthouse scores, visual appearance, build success) but I had not audited every file it touched. Had not checked .git/hooks/. I’d never opened that directory in my life. Had not read every package.json script before running npm install. I’d been doing exactly the thing the documentation warns about: treating the agent’s output as reviewed code when it was unreviewed code that I was about to execute on my machine.
It would be easy to read this as “Sandboxes are broken.” That’s not what I mean. The microVM does exactly what microVMs are supposed to do: it contains the consequences of arbitrary code execution behind a hardware boundary. What it cannot do is make the workspace contents safe, because the workspace contents are how the agent does its job. The agent has to be able to write files. You have to be able to read them. Shared region is necessary, and the shared region is where the threat model gets interesting.
Mitigation isn’t more isolation. The microVM is doing its job. Mitigation is discipline: treat the workspace contents the way you’d treat a pull request from a contributor you don’t know yet. Diff .git/hooks/ after agent sessions. Read package.json scripts before running npm install. Use the --branch flag, which creates a Git worktree so the agent works in an isolated branch you can review before merging. None of this is exotic. It’s just the practice of not treating autonomous-agent output as trusted code. Because it isn’t.
I’m spending this much space on it because it’s the part most people get wrong. Hypervisor boundary makes you feel safe, but you aren’t. Not completely. Both things have to be true at once for the product to work, and the Docker team built it that way on purpose. Good security architectures document their gaps and make sure the user knows what they’re signing up for.
What it actually costs
Hypervisor isolation isn’t free, and you can’t pretend otherwise. I tested this against my own production codebase, the same Astro blog I mentioned at the top, because synthetic benchmarks for sandboxed agent workloads don’t tell you much. You want to know what it feels like to do real work.
Caption: The same docker build --no-cache against the same Astro codebase. Host: 1:44.62. Sandbox microVM: 1:28.58. The isolation boundary is invisible to the workload. On this run, the sandbox actually finished faster.
I ran docker build --no-cache against the same Dockerfile and the same codebase, once on the host and once inside the sandbox. Host finished in 1:44.62. Sandbox finished in 1:28.58, actually faster, within noise across runs. The Docker Engine inside the sandbox is running on its own kernel with its own block device, completely isolated from the host, and the build doesn’t care. The microVM adds essentially zero overhead to the actual build.
One real-world caveat from running this on Apple Silicon: a Rust dependency in my Astro pipeline ships jemalloc that assumes 4K page sizes, which fails on sandbox VMs (16K pages). The build itself completed correctly. All 354 pages rendered, dist generated, but a teardown step exited non-zero. The fix was a one-line guard in the Dockerfile that checks for valid build output before exiting. Took 30 minutes to track down. Worth knowing about before you ship sandbox-aware Dockerfiles on Apple Silicon, because the symptom looks like a build failure when the build actually succeeded.
Verdict: for session-based agent work (a few hours on a project), the overhead disappears. For high-frequency sandbox creation (dozens per minute for short tasks), cold-start cost adds up. For the workload Sandboxes is designed for, which is giving an agent a real environment for a real session, the trade is sound.
Matching isolation to trust
Most discussions of containers versus VMs treat it as a binary, and that’s the wrong frame. The frame I’ve found useful, both for my own work and in conversations with engineering leaders who ask “do we really need microVMs for this?”, is a spectrum.
Caption: The Trust Spectrum. Match isolation strength to the trust profile of the workload.
On one end you have code you wrote yourself. Your team reviewed it, your CI tested it, your production runs it. A standard container is the right answer. Kernel is shared, daemon is shared, and none of that matters because the workload is known.
One step removed from that are CI/CD pipelines running your team’s code plus dependencies from registries you mostly trust. Mostly known, but the inputs are more variable. You add seccomp profiles, drop capabilities, write network policies.
Further along, supervised AI agents: tools that suggest code while a developer reviews each step. Human in the loop, so hardened containers with strict policies still work.
At the far end are autonomous AI agents. Nobody reviewing each command. Agents making decisions on your behalf, each one potentially different from the last. The trust profile isn’t “I trust this code” because there’s no fixed code to trust. It’s “I’m letting something operate on my system without supervision, and I want the failure mode to be ‘contained to a disposable VM’ rather than ‘on my laptop.'” That’s the workload that needs a microVM.
This is not a declaration that containers are obsolete. It’s the opposite. Containers are the right answer for everything on the left side of that spectrum, which is most of what runs in production today. MicroVMs extend the spectrum to the right, where containers were never going to be the right tool. The four isolation layers in Sandboxes (hypervisor, network, Docker Engine, credential proxy) are additive. They wrap containers in additional protection rather than replacing them. Inside every Sandbox is a microVM that runs containers. Containers haven’t gone anywhere, they’ve moved one level deeper in the trust stack.
“MicroVMs for AI agents, containers for everything else” is too crude. “Match the isolation to the trust profile of the workload” is the one that holds up.
Why everyone is converging here
Docker isn’t the only company that arrived at this answer, and the convergence tells you something.
Firecracker powers AWS Lambda and Fly.io’s microVM platform. gVisor intercepts syscalls in a user-space kernel. Kata Containers provides VM isolation behind a container-compatible interface. Modal runs serverless agent workloads on gVisor. E2B offers Firecracker-based sandboxes as a managed cloud service. Northflank ships Kata-based isolation for production AI workloads. All adopted at the same time, for the same reasons. Architecture everywhere looks the same: containers on the inside (because that’s how developers think), VM on the outside (because that’s where the boundary needs to be).
Docker Sandboxes is the local-first version. Most alternatives are cloud services where you pay per execution and your code runs on someone else’s machines. Docker put the same architecture on the developer’s laptop. CLI supports eight agents natively (Claude Code, Codex, Copilot, Gemini CLI, Kiro, OpenCode, Docker Agent, and Droid), plus a Shell mode for custom tooling. A standalone sbx CLI runs without Docker Desktop, so the architecture isn’t locked to a commercial product. MicroVM layer has an HTTP API that the open-source community has already started building on.
That’s a runtime. And Docker is positioning it to become the standard way to run autonomous coding agents, the way docker run became the standard way to run microservices ten years ago.
One more thing. Hardened Images and sandboxes address different layers of the same problem: Hardened Images for the supply chain (where binaries come from), sandboxes for runtime isolation (what those binaries can touch). Both exist because the assumption that “code from a trusted publisher is safe” stopped being reliable.
Looking back, looking forward
I’ve watched the industry rebuild its trust model three times in twenty years.
Bare metal to virtual machines, because we needed to put multiple workloads on the same hardware safely.
Virtual machines to containers, because we needed faster startup, lower overhead, and a packaging model that matched how developers actually ship code.
Now, containers to a different kind of virtual machine, because the workload changed and the kernel namespace stopped being enough. Not because containers were wrong, but because the new tenant needs more, and more looks like a hypervisor again.
Each of these transitions felt obvious in hindsight and contested at the time. I remember the arguments about whether containers were really secure enough for multi-tenant workloads. (They mostly weren’t, which is why we ended up with namespaced clusters and per-tenant VMs and gVisor and now microVMs for agents.) I expect the microVM argument to follow the same arc: contested for about a year, obvious within three.
My Astro migration taught me what it feels like to work alongside an autonomous agent that has real access to your system. More productive than doing it by hand, and more unsettling than I expected, once I realized how much I’d stopped tracking. Sandboxes don’t make the agent trustworthy. It just makes sure that when the agent does something you didn’t expect, the damage stays inside a box you can throw away. Workspace still requires your attention. Your skepticism. That combination (strong boundaries where you can enforce them, disciplined review where you can’t) is the model for working with autonomous code, and it’s probably going to stay that way for a while.
If you’ve been holding back on running AI coding agents because of permission prompts, accidental file changes, or just a feeling that something about the whole arrangement isn’t quite safe: that feeling was correct. Containers were the wrong fit for the workload. Sandboxes is the right one. Try it on a project you actually care about. That’s the only test that matters.
Get started with Docker Sandboxes →
View the full article
reporter

Foldable iPhone Reportedly Facing Mass Production Issues

Apple's first foldable iPhone is running into mass production yield problems at the pre-assembly stage, the leaker known as "Fixed Focus Digital" claims.


In a post today on Weibo, Fixed Focus Digital said Apple's troubles are not related to hinge reliability, as was previously reported, but rather due to surface-mount technology (SMT) during pre-assembly, with production yields failing to ramp up. The leaker framed the situation as somewhat concerning, stopping short of suggesting the fall launch is at risk.

The update arrives days after a separate leaker known as "Instant Digital" reported that the device's hinge was consistently failing to meet Apple's quality control standards under conditions of prolonged, high-frequency opening and closing. Instant Digital described that issue as one that "must be resolved with absolute perfection," though a follow-up post suggested the hinge difficulties were unlikely to affect the expected release window.

DigiTimes reported in April that production was already running roughly one to two months behind schedule, while still maintaining that a fall 2026 launch remained on track, with mass production planned to begin in July. Fixed Focus Digital reported in April that price negotiations with Apple's assembly partner were a potentially disruptive factor.

Whatever the precise nature of the problems, the picture that has emerged across multiple supply chain sources in recent weeks is one of unusual production difficulties. That said, a fall launch does not appear to be at risk; Bloomberg's Mark Gurman reported in April that the foldable iPhone remains on track for a September debut alongside the iPhone 18 Pro models, and that Apple is aiming to put it on sale at roughly the same time or slightly later. Gurman noted at the time, however, that "the release is six months away and production has yet to ramp up" and "the timing isn't final."

The foldable iPhone is expected to feature a 7.8-inch inner display and a 5.5-inch cover display, the A20 chip and C2 modem, a Touch ID power button instead of Face ID, and two rear cameras, with pricing rumored at around $2,000.Tags: Fixed Focus Digital, Foldable iPhone
This article, "Foldable iPhone Reportedly Facing Mass Production Issues" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

TrapDoor malware campaign puts developer workstations in CISO spotlight

A malicious package campaign across npm, PyPI, and Crates.io has put developer workstations back under scrutiny, after researchers said it targeted developer workflows and AI coding assistant files.
Researchers at Socket said the campaign, which they are tracking as TrapDoor, “spans more than 34 malicious packages and 384+ related versions and artifacts” across the three open-source ecosystems.
The packages were designed to steal developer secrets, including AWS credentials, GitHub tokens, SSH keys, browser data, environment variables, crypto wallets, and local development configuration files, according to Socket.
The findings indicate a bigger concern than just another malicious package incident. Developer environments increasingly sit at the intersection of source code, cloud infrastructure, CI/CD pipelines, AI coding tools, and privileged credentials. A compromise of one workstation can therefore give attackers a foothold beyond the developer’s machine.
The packages used execution points that are common in normal software development workflows. In npm, the malware relied on postinstall scripts. In PyPI, it used import-time execution to fetch and run remote JavaScript. In Crates.io, it abused Rust build scripts that execute during compilation. That makes the campaign harder to detect using controls focused on a single programming language or package registry.
TrapDoor also appears to reflect attackers’ growing interest in AI-assisted development environments. Socket said the campaign attempted to alter files used by AI coding tools, including .cursorrules and CLAUDE.md, using hidden Unicode instructions.
The apparent strategy was to trick AI assistants into running security-scan-like workflows that could lead to secret discovery and exfiltration.
That use of ordinary development mechanisms is what makes the campaign difficult to treat as a conventional malware incident, analysts said.
“TrapDoor represents a shift from opportunistic package abuse toward workflow-level compromise of developer environments,” said Sakshi Grover, senior research manager for IDC Asia Pacific Cybersecurity Services. “Earlier campaigns typically placed a malicious package, stole credentials on install, and moved on. TrapDoor is engineered around the full developer workflow, meaning the attack extends well beyond the package registry.”
Grover said the campaign’s cross-registry design makes it harder to spot from a single ecosystem view, since the malicious packages used the normal execution mechanisms of npm, PyPI, and Crates.io. The more serious concern, she said, is what happens after installation, when the malware attempts to persist on the developer machine and potentially use stolen SSH keys to move deeper into engineering systems.
“A single compromised workstation can quietly become an entry point into CI/CD pipelines and build infrastructure,” Grover said. “That’s not credential theft. That’s an initial access operation.”
Sanchit Vir Gogia, chief analyst at Greyhound Research, noted that the campaign is distinctive because it demonstrates an intimate understanding of how modern software is built.
“It does not stop at stealing credentials from one poisoned dependency,” Gogia said. “It targets the wider developer operating environment: package managers, AI coding assistants, Git hooks, shell profiles, SSH trust relationships, browser sessions, cloud credentials, CI/CD pathways, and the local workflow artifacts that developers and machines increasingly treat as legitimate context.”
Mitigation strategies
Gogia said the issue is no longer just endpoint security, but control over the systems and workflows that produce enterprise software.
“Developer environments must be treated as production-adjacent infrastructure,” Gogia said. “They carry code, secrets, identity, automation, cloud access, and now machine reasoning context. If an attacker owns the developer environment, they are not merely stealing a password. They are sitting beside the machinery that creates enterprise software.”
Mitigation starts with stronger controls around dependency installation and package behavior, according to Grover.
“Lockfiles alone don’t protect you,” she said. “You need automated scanning at install time against known-malicious packages and behavioral signals like unexpected postinstall scripts, remote payload fetching, or unusual network calls.”
Grover said least-privilege access for developer credentials is equally important, including scoped, short-lived keys and secrets management practices that avoid leaving credentials in environment variables or configuration files.
“If an attacker gets a key and it can’t move laterally, the campaign stalls,” she added.
Keith Prabhu, founder and CEO at Confidis, said CISOs should also prioritize hardened developer endpoints, package allowlisting, AI tooling governance, and zero-trust controls within local development environments.

View the full article
CSOonline

MFA Prompt Bombing: Why Your Second Factor Isn't Saving You

Multi-factor authentication (MFA) was supposed to close a critical gap in identity security. It meant that, even if an attacker possessed the account credentials, they couldn't log in without the second factor. While that logic was sound, attackers have now figured out that they don't need to steal the second factor: they just need the user to hand it over. If your workforce authenticates withView the full article
Hacker News

Stop treating AI governance as a review layer. Make it release infrastructure

I’ve spent years building compliance into security products. FedRAMP and Department of War Impact Level authorizations, vulnerability management pipelines: They all follow the same pattern. Build the product, then prove it meets requirements. The compliance layer sits outside the engineering workflow. It reviews what already exists.
That model worked when the product stayed static between audits. It breaks for AI.
AI systems change even when the base model does not. A retrieval index updates overnight. A new tool gets added to an agent’s action space. An evaluation that passed on Tuesday no longer reflects what the system does on Thursday. The compliance-as-review approach assumes that the thing you’re reviewing remains unchanged between review cycles. For AI, that assumption is fundamentally wrong. Most organizations I talk to are still trying to govern AI the way they govern traditional software: Build it, ship it, then ask legal to check the box. For AI, it leaves the release process blind to the thing most likely to change.
When I started researching how other countries handle this problem for my forthcoming book on China’s AI ecosystem, I found something that challenged my assumptions. Chinese AI companies don’t treat governance as a gate they pass after the model works. They treat it as release infrastructure: Compliance checkpoints embedded in the deployment pipeline itself. No checkpoint clearance, no product launch. The governance layer doesn’t review the product. It is part of the product.
In one AI deployment review I joined, the product team had everything the launch meeting usually rewards: Performance metrics, customer use cases, latency numbers and a firm release date. The missing pieces were not on anyone’s checklist. No one could point to a current, pipeline-generated record of the retrieval index feeding the model. No one owned the output-monitoring thresholds. No one had tied model evaluation results to an enforceable release gate. The team wasn’t ignoring governance. Governance simply had no place to live inside the actual release process.
The review layer is already failing
That scene is not unusual. When governance lives outside the engineering workflow, it competes with delivery timelines. Delivery timelines win every time. The NIST AI Risk Management Framework identifies govern, map, measure and manage as core functions for AI risk, but it doesn’t prescribe where those functions sit inside a release process. That leaves the hard architectural question to the security organization. Most companies default to what they know: A periodic review cycle borrowed from traditional IT compliance. That cycle was designed for systems that hold still between audits.
AI systems do not hold still. A model fine-tuned on last quarter’s customer data produces different outputs once this quarter’s data enters the pipeline. A retrieval-augmented generation system returns different answers depending on which documents sit in its index today versus yesterday. An agentic workflow that chains three models together produces emergent behaviors that no single-model evaluation captures. Governance-as-periodic-review was built for a world where the artifact under review doesn’t change. We are deploying artifacts that change continuously.
The gap between how fast AI systems evolve and how slowly review-layer governance cycles operate is the core vulnerability. Every week that gap widens, organizations accumulate governance debt they will eventually have to repay, either on their own terms or on a regulator’s.
What release infrastructure looks like in practice
When I researched China’s AI deployment process, I expected to find a heavy-handed approval system that slowed companies down. I found the opposite.
China requires companies deploying generative AI to complete a regulatory filing before their product reaches consumers. The filing demands documentation of training data sources, content safety mechanisms, output controls and user-facing disclosures. Companies that clear the process ship. Companies that do not, wait.
What surprised me was the speed. Baidu launched Ernie Bot to the public on August 31, 2023, sixteen days after China’s generative AI rules took effect. Dozens of companies followed within weeks. The filing process did not stop deployment. It sorted companies by those that had already built the evidence machinery to pass. The firms that treated compliance as a last-mile legal exercise fell behind.
That finding matters for Western security leaders. We should not replicate China’s regulatory model. The underlying operational problem, though, is identical. The EU AI Act reaches the same conclusion from a different regulatory tradition: Its conformity assessment and ongoing risk management requirements for high-risk AI systems assume continuous compliance, not one-time certification. The operational question both frameworks share is the same one I face in my own work: Where in the development process does governance actually live? If the answer is “after the model is trained and before it ships,” you’ve recreated the review-layer bottleneck. Engineering teams will find ways around it.
I saw the same pattern with SBOMs. When teams treated the SBOM as a document someone assembled for a customer questionnaire, it aged out almost immediately. When they generated it from the build pipeline, it became part of the product’s living operating record. Model documentation has to move the same way. A model card written by hand after release is a snapshot. A model card generated from the pipeline is evidence.
Three shifts security leaders should make now
I’ve started applying this principle in my own work and in how I advise teams evaluating AI deployment readiness. Three operational shifts make the difference.
First, move model documentation into the CI/CD pipeline. Treat model cards, training data provenance records and output behavior baselines the same way you treat SBOMs: As artifacts generated automatically during the build process, not as documents written by a compliance analyst after the fact. If your model documentation isn’t versioned alongside your code, it’s already out of date. Every model retraining cycle that doesn’t produce updated compliance artifacts widens your governance gap.
Second, make compliance evidence a deployment gate rather than a post-launch audit item. Your release pipeline probably already blocks deployment if unit tests fail or if a container image carries a critical vulnerability. Add AI governance checkpoints to that same pipeline. Does the model have a current risk evaluation against your organization’s defined thresholds? Is the training data lineage documented and traceable? Are output controls configured, tested and monitored? If the answer to any of those is no, the deployment doesn’t proceed. The pipeline already blocks vulnerable containers. AI governance checkpoints belong in the same layer. It’s extending your existing security architecture to cover a new class of risk.
The problem gets sharper when the AI system stops generating recommendations and starts taking actions. Third, treat agent identity as a first-class security control. As AI agents move into production environments, each one needs an identity in your IAM system with scoped permissions, audit trails and session-level accountability. An agent calling external APIs, reading customer data or triggering automated workflows is an actor in your environment. It requires the same identity governance you apply to human users and service accounts. I wrote about the identity and persistence challenges in stealthy ransomware operations earlier this year here at CSO. The same principles apply: If you can’t identify the actor, you can’t govern the action.
None of this requires waiting for regulation. The organizations that will be best positioned when AI-specific compliance mandates arrive, whether from the EU AI Act’s enforcement timeline, emerging US state-level legislation in Colorado and California or sector-specific rules from financial and healthcare regulators, are the ones building governance into their release infrastructure now. The ones still treating it as a review layer will scramble to retrofit what their competitors already ship with.
I learned that lesson by studying how a very different system solved this problem first. The regulatory traditions are different. The operational logic is the same: Governance that ships with the product beats governance that reviews it after the fact.
This article is published as part of the Foundry Expert Contributor Network.
Want to join?
View the full article
CSOonline

The CALMS Framework : Principles and Practices

Introduction
In my two decades of working with engineering teams, I have seen brilliant engineers struggle not because they lacked technical skill, but because they lacked a cohesive framework. We often talk about tools—Kubernetes, Jenkins, Terraform—as if they are the silver bullet for DevOps success. However, installing a tool is simple. Changing how an organization thinks, communicates, and delivers value is the real challenge.
Many organizations fail at DevOps because they view it as a project or a software installation. It is not. It is a fundamental shift in how IT organizations operate. This is where the CALMS framework becomes critical. Whether you are a student just entering the field or an IT manager tasked with transforming your department, understanding CALMS is the difference between implementing DevOps and actually living it.
If you are looking to master these principles, resources like DevOpsSchool provide structured learning paths that go beyond the theory and into real-world application. Understanding the CALMS framework allows you to diagnose why your pipelines are stalling, why silos exist between your development and operations teams, and how to start moving toward a more mature, reliable, and collaborative engineering culture. Let us break down exactly what this means and how you can apply it today.
What Is the CALMS Framework?
The CALMS framework is a mental model used to assess and improve an organization’s DevOps maturity. It was coined by Jez Humble, one of the authors of the seminal book The DevOps Handbook.
In simple terms, CALMS is a diagnostic tool. If you look at your organization and feel that things are slow, error-prone, or disconnected, you can hold your current process up against the five pillars of CALMS to see exactly where the breakdown is occurring. It shifts the focus away from “which tools are we using” toward “how are we working.”
It originated from the realization that DevOps is not a product you can buy; it is a philosophy. By assessing your organization against these five pillars, you can map out a transformation journey that is sustainable and scalable.
What Does CALMS Stand For?
The acronym CALMS stands for Culture, Automation, Lean, Measurement, and Sharing. Each component represents a critical dimension of a high-performing DevOps team.
LetterMeaningPurposeReal-World ExampleCCultureCreates a shared mindsetBlameless post-mortems after an outageAAutomationReduces manual toilUsing CI/CD pipelines to deploy codeLLeanMinimizes wasteBreaking large features into small releasesMMeasurementTracks progressMonitoring error rates and deployment speedSSharingBreaks down silosInternal wiki for tribal knowledge Why the CALMS Framework Matters in DevOps
Without a framework like CALMS, DevOps initiatives become random acts of automation. You might have a great CI/CD pipeline, but if your culture is one of fear and blame, developers will still be afraid to deploy frequently.
CALMS matters because it addresses the systemic nature of software delivery. It ensures that you are not just optimizing one part of the pipeline while ignoring the others. It promotes a holistic view of the software development lifecycle, ensuring that engineering, operations, and business goals are aligned.
C = Culture Explained Simply
Culture is the most important, and often the hardest, part of the CALMS framework. It is the “software” that runs on the “hardware” of your team.
Breaking Silos
In traditional IT, developers wrote code and threw it over the wall to operations. The operations team, responsible for stability, would then be incentivized to block changes. Culture in CALMS is about breaking that wall. It promotes shared ownership: “You build it, you run it.”
Shared Accountability
When a system fails, a healthy culture does not look for a person to punish. Instead, it looks for the process that allowed the failure to happen. This is the hallmark of a high-trust, high-performance engineering environment.
A = Automation Explained Simply
Automation is the engine of DevOps. However, automation for the sake of automation is a trap. You should only automate processes that are stable and repeatable.
Reducing Toil
Manual, repetitive tasks are a drain on engineering morale and a source of human error. Automation allows you to shift your engineers’ focus from “keeping the lights on” to “building new features.”
Essential Tools
CI/CD: Jenkins, GitHub Actions, GitLab CI. Infrastructure as Code (IaC): Terraform, Ansible. Orchestration: Kubernetes, OpenShift. Workflow Example: Instead of manually configuring a server, use Terraform to provision the environment. This ensures the infrastructure is consistent, version-controlled, and reproducible.
L = Lean Explained Simply
Lean comes from the manufacturing world, specifically Toyota’s Production System. In DevOps, Lean is about optimizing the flow of value from the developer’s laptop to the customer’s screen.
Removing Waste
Waste in software development includes waiting for approvals, context switching, and large, infrequent releases that cause merge conflicts. Lean encourages smaller, more frequent batches. If you deploy ten small changes a day, a bug is easier to isolate than if you deploy one massive change once a month.
Faster Feedback
By shortening the time between a code commit and the user seeing the result, you get faster feedback. This allows you to pivot quickly if the feature is not working as intended.
M = Measurement Explained Simply
You cannot improve what you do not measure. Measurement in CALMS is about visibility and data-driven decision-making.
Key Metrics
Deployment Frequency: How often do you ship code? Lead Time for Changes: How long does it take from commit to production? Mean Time to Recovery (MTTR): How fast can you fix a system when it breaks? Measurement keeps the team honest. It helps you move away from gut feelings and toward objective reality. When you have a dashboard showing your metrics, the conversation shifts from “I think we are doing well” to “Our deployment frequency has increased by 20 percent this month.”
S = Sharing Explained Simply
Sharing is the glue that holds the framework together. It involves the free flow of information across the organization.
Knowledge Management
If only one person knows how a critical system works, you have a single point of failure. Sharing involves creating a culture of documentation, regular brown-bag sessions, and cross-team rotations.
Collaboration
Sharing is not just about writing documentation; it is about cross-functional collaboration. When developers, operations, and security teams work in the same Slack channels or shared project management boards, they share the context, the constraints, and the success.
Real-World Example of CALMS in Action
Imagine an e-commerce company struggling with slow deployments and frequent outages. They decide to adopt the CALMS framework:
Culture: Management shifts from individual performance metrics to team-based goals. They implement blameless post-mortems. Automation: They implement a CI/CD pipeline. Every code commit automatically triggers tests and deploys to a staging environment. Lean: The team stops doing bi-monthly “big bang” releases. They move to daily releases of small, incremental features. Measurement: They set up a dashboard to track deployment frequency and incident rates. Sharing: The team starts a “knowledge share” session every Friday to discuss technical challenges and successful fixes. Within six months, the team is deploying daily, outages are resolved faster, and the engineers are happier.
CALMS Framework vs Traditional IT Practices
Traditional ITCALMS FrameworkSiloed teams (Dev vs. Ops)Cross-functional collaborationManual, error-prone processesHigh levels of automationLarge, infrequent releasesSmall, continuous releasesFear of failure/Blame cultureLearning from failure/Psychological safetyReactive monitoringProactive, data-driven observability Benefits of the CALMS Framework
Better Team Collaboration: Reduces friction and finger-pointing between departments. Faster Releases: Automation and Lean principles accelerate the path to production. Higher Reliability: Measurement and automation lead to more stable systems. Reduced Downtime: Faster recovery times thanks to better visibility and automated rollback capabilities. Improved Customer Experience: Users get new features and fixes faster. Common Challenges While Implementing CALMS
Resistance to Change: People are naturally comfortable with the status quo. Start small and demonstrate quick wins. Lack of Automation Knowledge: Teams may not know where to start. Invest in training and upskilling. Poor Communication: Silos are hard to break. Start with cross-team meetings or shared project boards. Weak Measurement Systems: You cannot measure what you do not track. Start with one simple metric, like deployment frequency. How Beginners Can Apply CALMS Principles
If you are new to the field, start here:
Step 1: Learn Collaboration: Offer to sit in on a meeting with another team. Learn their language. Step 2: Practice Automation: Automate a single, annoying manual task. Write a script to back up a file or verify a system state. Step 3: Understand Lean Thinking: Look for “waiting” time in your own work. What is blocking you? Step 4: Learn Monitoring Basics: Install a simple monitoring tool on your personal projects. Look at the graphs. Step 5: Practice Knowledge Sharing: Write down how you solved a specific technical problem and share it with your peer group. Role of CALMS in Modern Cloud-Native Engineering
In the age of Kubernetes and microservices, CALMS is more important than ever. Cloud-native systems are inherently complex. Without Culture, teams will not communicate the complexity. Without Automation, you cannot manage hundreds of containers. Without Lean, you will drown in configuration debt. Without Measurement, you will not know which service is causing the latency. And without Sharing, the complexity of the microservices architecture will be impossible to maintain.
Role of DevSecOps in the CALMS Framework
DevSecOps is the natural extension of CALMS. It integrates security into every step:
Culture: Security becomes everyone’s responsibility, not just the “security team’s” problem. Automation: Automated security scanning is built into the CI/CD pipeline. Lean: Security checks are performed early and often, not as a gate at the end. Measurement: Security vulnerabilities are tracked just like bugs. Sharing: Security knowledge is shared across development and operations teams. Common Beginner Mistakes
Focusing Only on Tools: Buying expensive software without fixing the process is a waste of money. Ignoring Collaboration: You can automate everything, but if you don’t talk to each other, you will automate the wrong things. Weak Monitoring Knowledge: Guessing at what is wrong during an outage is a recipe for disaster. Avoiding Documentation: “The code is the documentation” is a myth. Write down how things work. Best Practices for Applying the CALMS Framework
Encourage Collaboration: Create environments where people feel safe to suggest changes. Automate Repetitive Work: If you do it more than twice, automate it. Measure Everything Important: Focus on metrics that impact the customer and the team’s efficiency. Share Learning Regularly: Celebrate failures as learning opportunities. Continuously Improve Workflows: Never settle for “good enough.” Small, daily improvements compound over time. Industries Benefiting from the CALMS Framework
Banking & Finance: Benefits from improved reliability, auditability, and regulatory compliance through automation and measurement. Healthcare: Leverages CALMS for faster, safer delivery of critical patient-care applications and secure data management. E-Commerce: Uses the framework to handle high traffic spikes and deploy updates rapidly without downtime. SaaS Platforms: Relies on CALMS to maintain uptime and innovate faster than competitors. Telecom: Uses automation and Lean principles to manage massive, complex network infrastructures. Enterprise IT: Adopts CALMS to modernize legacy systems and break down rigid internal silos. Career Opportunities for Professionals Understanding CALMS
Organizations are actively seeking professionals who understand the why behind DevOps, not just the how.
DevOps Engineer: Focuses on the “Automation” and “Measurement” aspects. Cloud Engineer: Leverages CALMS to build scalable, automated cloud infrastructures. SRE (Site Reliability Engineer): Deeply involved in “Measurement” (reliability) and “Sharing” (post-mortems). Platform Engineer: Builds internal tools that enable other teams, effectively “Sharing” infrastructure capabilities. DevSecOps Engineer: Integrates security into the CALMS framework. Industry demand for these roles is surging because companies know that a tool-only approach to DevOps is a losing strategy.
Certifications & Learning Paths
To implement CALMS effectively, you need hands-on skills. The learning ecosystem at DevOpsSchool is designed to bridge the gap between theoretical knowledge and practical execution.
CertificationBest ForSkill LevelFocus AreaDevOps FoundationBeginnersEntryCulture & PrinciplesCI/CD Pipeline EngineerIntermediateIntermediateAutomationCloud-Native ArchitectAdvancedExpertAutomation & ArchitectureSRE ProfessionalAdvancedExpertMeasurement & Reliability Future of the CALMS Framework
The CALMS framework is evolving. We are seeing:
AI-Assisted Automation: AI is helping us write code, optimize configurations, and even predict outages. Platform Engineering: A shift toward treating internal developer platforms as products (perfectly aligned with the Sharing principle). DevSecOps Maturity: Security is becoming truly “invisible” in the pipeline. GitOps Adoption: Using Git as the single source of truth for infrastructure, which is a perfect intersection of Culture, Automation, and Sharing. FAQs
1. What is the CALMS Framework? It is a mental model that helps teams assess and improve their DevOps maturity by focusing on Culture, Automation, Lean, Measurement, and Sharing.
2. What does CALMS stand for? Culture, Automation, Lean, Measurement, and Sharing.
3. Why is CALMS important in DevOps? It moves the focus from tools to the human and process elements of DevOps, ensuring that transformation efforts are holistic and sustainable.
4. Is automation mandatory in CALMS? Automation is essential for speed and reliability, but it is just one of the five pillars. It must be balanced with culture and measurement.
5. What is Lean in CALMS? Lean is about eliminating waste and optimizing the flow of work to deliver value to the customer faster.
6. Why is measurement important? Measurement provides the data needed to make informed decisions and proves that your improvements are actually working.
7. Can beginners learn CALMS? Absolutely. It is a mindset that can be applied at any stage of your career, from intern to architect.
8. Is CALMS useful in DevSecOps? Yes, it is fundamental to DevSecOps, as it emphasizes the culture of shared security and automated compliance.
9. How do I start with Culture? Start by fostering psychological safety. Encourage teams to discuss failures openly without fear of retribution.
10. What is the role of Sharing? Sharing ensures that information, best practices, and lessons learned are distributed across the organization, preventing knowledge silos.
11. Is CALMS a replacement for Agile? No, it is complementary. Agile focuses on how work is planned and managed, while CALMS focuses on how it is delivered and operated.
12. How often should we measure? Measurement should be continuous. Use automated dashboards to get real-time insights into your performance metrics.
13. Does CALMS apply to small startups? Yes. Even a two-person team can benefit from Lean principles and a culture of shared ownership.
14. What if management resists CALMS? Show them the data. Start small, get a win, measure the improvement, and present the results to leadership.
15. Can I use CALMS in a non-tech team? The principles of collaboration, removing waste, and measuring success are applicable to almost any business function, from marketing to HR.
Final Thoughts
The CALMS framework is not just another corporate acronym. It is a reality check for the modern engineering team. In my experience, the biggest bottleneck in DevOps is never the code; it is the friction between people and the lack of clarity in our processes.
If you focus on Culture, you will create a team that wants to innovate. If you embrace Automation, you will free up time for that innovation. If you apply Lean, you will deliver value faster. If you leverage Measurement, you will know exactly how you are performing. And if you practice Sharing, you will build an organization that learns and grows together.
Do not try to force all five pillars at once. DevOps is a journey of continuous improvement. Start by looking at your team’s current friction points and pick one CALMS pillar to focus on this month. You will be surprised at how quickly small, intentional changes can transform your delivery capability.
View the full article
reporter

Iranian Hackers Deploy MiniFast and MiniJunk V2 via Phishing and SEO Poisoning

The Iranian state-sponsored threat actor known as Nimbus Manticore (aka Screening Serpens and UNC1549) has been attributed to a fresh campaign using lures impersonating organizations in the aviation and software sectors across the U.S., Europe, and the Middle East following the joint U.S.-Israeli military campaign against the country in late February 2026. The activity, besides embracingView the full article
Hacker News

Vulnerabilities have become cyber attackers’ No. 1 door to the enterprise

Patching practices are coming under intense pressure of late, as time-to-exploit windows accelerate — a new reality likely to worsen as AI assistance in attack chains rises.
Now cyber defenders have another cause for flaw alarm: Vulnerability exploitation has significantly pulled away from stolen credentials as the most common entry point in security breaches, according to the latest edition of Verizon’s annual Data Breach Investigations Report (DBIR).
Verizon researchers found that exploited flaws were the root cause of breaches in 31% of cases, with credential abuse blamed for 13% of security failures. In a nod to patch management difficulties in the enterprise, only one in four (26%) critical vulnerabilities were fully remediated in 2025 with the median patch time rising to 43 days, up from 32 days the year prior, according to Verizon’s DBIR.
Root cause analysis
Verizon’s study is based on an analysis of 31,000 security incidents — of which 22,000 were confirmed data breaches — involving victims spanning 145 countries.
Incident response experts quizzed by CSO confirmed the rise in vulnerability exploitation as a means for breaking into enterprises is real.
“Attackers follow the path of least effort at scale, and right now that path runs through unpatched perimeter and edge devices, where a working exploit needs no prior access, no phished user, and no breach data to buy,” notes Daniel Bechenea, security manager at offensive security and vulnerability assessment platform Pentest-Tools.com.
Bechenea argues that exploitation has overtaken credential abuse because the patching of known exploits is failing to keep up with the rise of critical vulnerabilities.
Chris Wysopal, co-founder and chief security evangelist at Veracode, agrees.
“Organizations are still simply not fixing flaws fast enough,” he says.
According to Verizon’s analysis, only about 26% of CISA Known Exploited Vulnerabilities (KEVs) were fully remediated in 2025, down from 38% the prior year. Meanwhile, the volume of critical-severity vulnerabilities organizations had to patch grew by 50% year-on-year.
James John, an incident response manager at Bridewell, offered a contrasting perspective on the relative importance of vulnerability exploitation and credential abuse over the full lifecycle of security breaches.
“We’re still seeing identity is the primary chokepoint,” says John, whose cybersecurity services and incident response firm contributed data to the Verizon report. “Exploitation may now win the race to the front door, but stolen credentials are still the thread running through most intrusions we respond to; they’re just used later in the attack, to move laterally and reach the data that matters.”
The Verizon report also attributed 16% of initial breach access to phishing, par with the year prior, and 6% to pretexting, which the researchers noted has become more common in ransomware and extortion attacks.
That latter point somewhat muddies the report’s credentials conclusion, John notes.
“Some of the apparent decline [in credential abuse] is also measurement rather than reality, as credential theft and pretexting blur together,” he tells CSO.
As companies rely more heavily on external vendors, threat actors are targeting the extended supply chain as well, with breaches involving a third party now accounting for 48% of all security incidents covered by Verizon’s DBIR.
Verizon’s DBIR — now in its 19th year — combines real-world incident and breach casework from law enforcement, forensic firms, and cyber industry sharing groups such as national CERTs, along with data from Verizon’s work with its own clients. Findings from what’s regarded as the industry’s benchmark study on data breaches are supported by recent broadly comparable studies.
Google Cloud Security’s latest Cloud Threat Horizons Report, for example, also found that attackers are pivoting toward exploiting unpatched third-party software vulnerabilities rather than relying primarily on stolen or weak credentials.
Software vulnerabilities became the biggest single initial access vector (44.5% of incidents), overtaking credential abuse, according to the Google Cloud study.
AI already adding to the threat landscape
Although the latest DBIR report uses 2025 data — predating the latest frontier AI security model advancements such as Anthropic’s Mythos — greater reliance by cybercriminals on AI still emerges from detailed post-mortems on security breaches.
“AI is being leveraged by threat actors to accelerate the time to exploit known vulnerabilities, shrinking the window for defence from months to mere hours,” Verizon warned.
Last week the Google Threat Intelligence Group (GTIG) released evidence of a zero-day exploit developed by a cybercriminal group with the help of AI.
Breach remediation strategies need to change
Muhammad Yahya Patel, vCISO and cybersecurity advisor for EMEA at managed security services vendor Huntress, says CISOs need to rapidly improve their vulnerability management and identity security in light of the Verizon DBIR findings.
“Vulnerability exploitation, credential theft, multi-channel social engineering, and supply chain compromise are all being deployed at scale simultaneously,” Patel says. “The organizations best positioned are those that have built defense in depth across all of these vectors.”
Patel adds: “More organizations need to shift their vulnerability management program to a risk-based, continuous [approach], tied to real-time exploitation intelligence — not scheduled patch cycles that leave exploitation windows wide open for days and weeks.”
Raghu Nandakumara, VP of industry strategy at microsegmentation and breach containment vendor Illumio, argues that even though more vulnerabilities are being fixed as enterprise patching practices improve, the backlog of flaws requiring remediation is still growing faster than security teams can keep up.
“The spike [in vulnerability instances] has been driven by a convergence of forces, including more AI-assisted discovery, greater reliance on third-party and open-source code, a growing number of connected systems, and a disclosure ecosystem that’s now far more active and incentivized than it was even a few years ago,” Nandakumara says.
Ransomware payments declining but threat remains potent
Ransomware was a feature in nearly half of all breaches (48%) covered by the DBIR, up from 44% the year prior, even though ransom payments have declined (69% of victims did not pay).
Aparna Rayasam, CEO of network security firm Atsign, says that this shift in payment rates is spurring ransomware to evolve toward a different business model.
“Because victims aren’t paying for decryption keys anymore, attackers have shifted heavily toward data exfiltration and extortion,” he says. “Attackers are compensating for smaller individual payouts by executing a higher volume of cheaper, automated attacks.”
Rayasam adds: “Use of AI makes this model even more lucrative for the ransomware attackers.”
Bridewell’s John offered a contrasting perspective, arguing that although ransomware attackers are no less successful in attacking enterprises, they are finding it more difficult to extract payment from victims.
“The drop [in ransomware payments] reflects genuine progress and not attackers losing their edge,” John tells CSO. “More organizations have tested backups and rehearsed recovery, so they can credibly refuse to pay, and the DBIR notes refusals are rising even in cases involving encryption, not just data theft.”
This reduction in payment rates means that attackers are becoming more aggressive in their attempts to disrupt a business in order to pile greater pressure on them to pay.
For example, UK high street retailer Marks & Spencer suffered weeks of outages and millions in losses as the result of a ransomware attack.
“The leverage is shifting from ‘we have your data’ to ‘we can keep you offline,’ which matters far more when downtime affects essential services,” John concludes.
View the full article
CSOonline

KnowledgeDeliver LMS Flaw Exploited to Deploy Godzilla and Cobalt Strike

A now-patched high-severity security flaw affecting Digital Knowledge KnowledgeDeliver, a Learning Management System (LMS) popular in Japan, was exploited as a zero-day to deliver the Godzilla web shell and ultimately facilitate the deployment of Cobalt Strike Beacon. The vulnerability, tracked as CVE-2026-5426 (CVSS score: 7.5), stems from the use of hard-coded ASP.NET machine keys, leading toView the full article
Hacker News

Security experts caution MFA alone can no longer stop threat actors

Cybersecurity experts are warning enterprise admins about an increasing number of phishing campaigns aimed at stealing Microsoft 365 (M365) access tokens to bypass multifactor authentication login protection.
Phishing kits aimed at capturing M365 tokens aren’t new; some reports say these kits have been around since 2021. One of the latest is EvilTokens, which researchers at Sekoia say has been circulating since February. And earlier this month, Microsoft also issued a warning about other adversary-in-the middle phishing schemes that steal authentication tokens, and, separately, about campaigns that exploit OAuth protocol functionality to manipulate URL redirection to bypass conventional phishing defenses.
Lowers the barrier to entry
But, said the US Federal Bureau of Investigation (FBI) in a warning last week, the new Kali365 phishing-as-a-service platform “lowers the barrier of entry, providing less technical attackers access to AI-generated phishing lures, automated campaign templates, real-time targeted individual/entity tracking dashboards, and OAuth token capture capabilities.”
It’s increasingly being leveraged by threat actors. On April 24, for example, security vendor Arctic Wolf said that it had detected a large-scale device code phishing campaign impacting organizations that was run by a threat actor using the Kali365 service. Four days later, researchers at Gurucul issued a similar warning, adding the new Kali365 kit “is rapidly becoming a preferred weapon” of threat actors. Both Kali365 and EvilTokens platforms trick employees into entering a code on a legitimate Microsoft login page that allows attackers to steal OAuth tokens.
But, Gurucul warned CSOs, “[Kali365] signals a shift toward highly professionalized attack models.” The researchers noted, “This is not just a single hacker working in isolation. Instead, it is a full-scale commercial operation. It is designed to lower the barrier for entry for criminals globally. By providing a ready-made infrastructure for deception, this kit places sophisticated capabilities in the hands of novice attackers.”
Move beyond MFA as a ‘checklist item’
CSOs should take the warnings as a reminder that phishing detection lessons are an essential part of security awareness training for all employees.
The FBI caution “[also] reminds us that multifactor authentication is no longer the single step that must be present for protection,” said Robert Beggs, CEO of Canadian incident response firm Digital Defence.
“Organizations have to move beyond having it as a ‘checklist item’ and instead focus on a defense in depth approach. Organizations have to block or tightly restrict Microsoft’s OAuth device code authentication flow using Conditional Access. Additional controls include revoking OAuth tokens proactively, monitoring for unauthorized device registrations, and monitoring to detect new or malicious inbox rules.”
Professional attack model
The Kali365 service provides templates, management dashboards, and integrated tools that lower the skill barrier for implementing large-scale attacks to the threat actor subscribing to it. Subscriptions start at $250 for 30 days and go up to $2,000 for 365 days.
Once signed up, Arctic Wolf said, Kali365 affiliates can rapidly generate branded phishing lures impersonating common enterprise services such as Adobe Acrobat Sign, DocuSign, and SharePoint. The service includes a modular lure‑generation system that allows threat actors to produce hundreds of distinct variants by mixing language localization, presentation layout, Microsoft‑ecosystem impersonations, and multiple document formats in English, Spanish, French, German, Portuguese, Italian, Dutch, Japanese, Korean, Chinese, Arabic, Turkish, Polish, and Russian.
Beggs noted that the use of AI generated phishing lures, assuming the AI has been properly trained against the client business and supplied with the correct cultural contexts, results in trustworthy-appearing documents that are difficult to identify and block in a large-scale attack.
Subscribers can take advantage of eight hard-coded email templates, with subject lines like “Voicemail from [with room for a name]”, “Signature Required,” “Invoice #INV,”, “Document Shared,” and “Account notification for [with room for an email address].”
Arctic Wolf said it has also seen cases where, after gaining initial access, the threat actor created malicious inbox rules within Microsoft 365, configuring rules that automatically moved emails containing keywords such as “spam,” “phish,” “click,” “link,” and “SharePoint” to a separate folder and marked them as read. This behavior effectively suppressed security-related notifications and warnings to the user, enabling the threat actor to maintain access while reducing the likelihood of detection.
In device code mode, victims are redirected to an obfuscated landing page that is designed to only render in a real browser session. Upon page load, the Kali365 backend dynamically generates a legitimate Microsoft OAuth device code.
According to the FBI, the attack then works like many other phishing scams: An attacker sends a phishing email with a message that includes a link to a legitimate Microsoft verification page, and instructions to enter the generated code. This code authorizes the attacker’s device to access the victim’s account. The Kali365 backend then captures OAuth access and refresh tokens, giving the threat actor access to the targeted individual’s/entity’s Microsoft 365 account, including Outlook, Teams, and OneDrive, until the compromise is detected and the tokens revoked. Using those tokens, the attacker doesn’t need to enter a password or complete any additional MFA challenges.
In some cases, Arctic Wolf added, following token acquisition, the threat actor would use the authenticated session to register an additional device within the victim’s Microsoft environment. This step extended access beyond the initial token by establishing a trusted device association tied to the compromised account.
Mitigation
In its alert, the FBI urged Microsoft 365 admins to restrict device code flow, since limiting or blocking device authentication codes can help prevent or minimize this style of attack. They should also create conditional access policies to block device code flow for all users, with limited exceptions for required business processes; audit existing device code flow usage to identify legitimate dependencies before creating a conditional access policy; and block authentication transfer policies to prevent users from transferring authentication from computers to mobile devices.
If an admin cannot completely restrict device code flow usage, the FBI says they should exclude emergency access accounts to prevent lockouts.
Christopher Kayser, CEO at Cybercrime Analytics and author of the book Cybercrime Through Social Engineering, said IT departments must find ways to reinforce to employees that they should not be quick to click on communications that seem unusual or potentially fraudulent. And it’s not just ordinary employees who can be hit by phishing scams, he pointed out. Higher levels of management with authority to transfer funds are targeted by business email compromise (BEC) scams.
Typically, he added, when signing into M365, users aren’t asked to input a code; they should be reminded that an email that asks for a code should be a red flag that triggers a call to the IT department.
Identity-centric security is key
Fritz Jean-Louis, principal cybersecurity advisor at Info-Tech Research Group, said defenders should shift to identity-centric security and treat phishing primarily as an identity compromise risk.
This not only means enforcing phishing-resistant MFA through passkeys or other FIDO2 approved login measures, but also strengthening session controls, and monitoring for anomalous authentication behavior, including token misuse and suspicious OAuth activity.
Admins should also adopt continuous access evaluation, moving beyond point-in-time authentication by dynamically assessing user and device risk throughout active sessions, enabling real-time response to evolving threats.
In addition, responders should leverage behavioral signals by measuring activity and encouraging users to report suspect behavior, and incorporating human telemetry, such reporting speed and interaction patterns, into detection strategies.
Jean-Louis said admins also need to reduce their organization’s blast radius by implementing stronger outbound monitoring, automated containment triggers, and tighter controls on account misuse, to limit lateral spread.
Finally, he recommended that admins segment high-risk users and functions by applying enhanced security controls and providing isolated environments for executives, finance, and privileged IT roles.
View the full article
CSOonline

Ferrari Reveals $640,000 EV Co-Designed by Jony Ive

Ferrari today unveiled the Luce, its first fully electric car designed with help from Apple's former design chief Jony Ive.


"Designed with Sir Jony Ive and Marc Newson at the creative collective LoveFrom, a singular design language unites the exterior, interior, and interface with clarity and refined simplicity throughout," said Ferrari.

The exterior of the car has a "smooth, continuous, and uninterrupted" design, with a "shell-like form" and "floating front and rear aerodynamic wings."


The interior has "precision-engineered mechanical buttons, dials, toggles, and switches" combined with "multifunctional digital displays."

The three-spoke steering wheel is machined from 100% recycled aluminum.


The four-door, five-seat Luce is powered by four electric motors providing up to 1,035 horsepower, and it is equipped with a high‑capacity 122 kWh battery. Ferrari says the car can accelerate from 0-100 km/h (0-62 mph) in just 2.5 seconds.

A dedicated app offers climate controls and charging settings, and it displays the car's status.


Luce pricing starts at €550,000 ($640,000) in Europe, with production set to begin in late 2026. The car will launch in the U.S. in the second quarter of 2027.

Tags: Ferrari, Jony Ive, LoveFrom
This article, "Ferrari Reveals $640,000 EV Co-Designed by Jony Ive" first appeared on MacRumors.com

Discuss this article in our forums

View the full article
reporter

Project Glasswing has uncovered 10,000 vulnerabilities: Anthropic

Anthropic says it and upwards of 50 partners involved in Project Glasswing have uncovered an estimated 10,000 critical or high-severity vulnerabilities in their software offerings.
The company launched the cybersecurity initiative, which is built around Claude Mythos Preview, in April, stating that its launch partners would use it as part of their defensive security work.
Anthropic said it created Project Glasswing when capabilities in its new frontier model “revealed a stark fact: AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.” At the time, it also indicated that it was committing upwards of $100 million in usage credits, as well as an additional $4 million in donations to open source security organizations.
In an update published late last week, Anthropic stated, “for the last few months we have used Mythos Preview to scan more than 1,000 open-source projects, which collectively underpin much of the internet — and much of our own infrastructure.”
During that process, Mythos Preview found 6,202 high or critical severity vulnerabilities in these projects, 1,752 of which have since been assessed by six independent security research firms.
Maintainers facing bug report deluge
Of these, Anthropic stated, 90.6% (1,587) “have proved to be valid true positives, and 62.4% (1,094) were confirmed as either high or critical severity. That means that even if Mythos Preview finds no further vulnerabilities, at our current post-triage true-positive rates, it’s on track to have surfaced nearly 3,900 high or critical severity vulnerabilities in open source code — in addition to those it has found for Project Glasswing’s partners.”
Authors of the report noted that, on top of the regular challenges of maintaining open-source software, “maintainers have been facing a deluge of low-quality, AI-generated bug reports. Indeed, several maintainers have told us they’re currently severely capacity constrained, and some have even asked us to slow down our rate of our disclosures because they need more time to design patches.”
Anthropic estimated that it has disclosed 530 high or critical severity bugs to maintainers so far, and is aiming to disclose another 827. Of those 530 bugs, 75 have been patched, and there have been 65 public advisories. The company said that the relatively low number is due to three factors: first, the 90 day window set out in its coordinated vulnerability disclosure policy has not closed, second, it is probably undercounting because some flaws have been patched without disclosure, and finally, the security ecosystem is already overloaded.
“The relative ease of finding vulnerabilities compared with the difficulty of fixing them amounts to a major challenge for cybersecurity,” the report’s authors noted.
Nonetheless, Glasswing’s success so far has led the company to release Claude Security in beta for its enterprise customers, and it has begun its Cyber Verification Program to allow legitimate security pros to use its models in their work without certain of its safeguards.
Mark Tauschek, distinguished analyst at Info-Tech Research Group, said Anthropic’s decision to keep access to Claude Mythos Preview restricted through Project Glasswing is one of the clearest signals yet that frontier AI capabilities have crossed a real threshold in cybersecurity.
The company, he said, “deserves some credit for the transparency of its system card and the structure of Project Glasswing. But being transparent about the problem is not the same thing as solving it.”
According to Tauschek, “the update validates the practical reality that IT and security leaders now have to deal with the fact that the cost of discovering software vulnerabilities has dropped dramatically. If a single AI model can surface thousands of serious vulnerabilities across foundational software in a matter of weeks, the window between vulnerability discovery and exploitation will keep compressing.”
Organizations still treating patching as a quarterly exercise are operating with materially more risk than they were even a short time ago, he added.
Tauschek said that the fact that some maintainers have asked Anthropic to slow down should not be seen as resistance to better security. “Rather, it points to a capacity problem that has been building for years,” he said. “Many of the open-source projects enterprises rely on are maintained by small teams or volunteers, often people with day jobs.”
A key bottleneck has moved
Meanwhile, he said, the “organizations depending on that code operate at a massive scale. AI can accelerate discovery, but it does not create the human capacity required to validate findings, design safe patches, test them, and get them deployed. This also forces a rethink of defense-in-depth.”
Kellman Meghu, CTO of DeepCove Cybersecurity, added that nothing in the Project Glasswing update is surprising to him. “Our company had figured out almost two years ago that, in the hands of a competent researcher, the ability [of AI] to find vulnerabilities and exploit them were greatly accelerated,” he noted. “I think the change now is that the barrier of entry to drive the prompts in a large language model has dropped significantly. This will only get better, and is our new reality.”
DeepCove, he said, “has had to accelerate its patching and controls assessment, which now includes leveraging large language models to help identify and patch or build compensating controls for our services and our customer infrastructures.”
According to Meghu, “finding bugs is now cheap, but patching them is still slow and human-bound in many cases. Clients have change management processes, regulatory testing windows, and change blackouts that make absorbing this pace genuinely hard.”
What Anthropic’s update really shows, he pointed out, “is that the bottleneck in cybersecurity has moved from finding vulnerabilities to absorbing patches and adapting client defenses fast enough to keep up.”
His take echoes that of Anthropic, which noted, “the bottleneck in fixing bugs like these is the human capacity to triage, report, and design and deploy patches for them.”
The operational pressure of the new patch cadence is “as immediate as the offensive threat,” Meghu said. “We’ve responded by building AI-assisted auditing into our own development pipeline and tightening client patch SLAs on critical dependencies. But this is not an easy process to manage. We do not blindly trust LLMs or agents to operate autonomously, and this has resulted in significant change in operator assisted processes for LLM integration.”
As well, noted David Shipley, CEO of Beauceron Security, “the headline grabber that everyone’s been paying attention to is the 10,000 potential vulnerabilities it found, and then, of those, 6,000 being critical, but when you actually pare the numbers down, you get closer to 1,500 that are actually human verified, legitimate, and so that’s quite a fall off.”
“[Anthropic also stated that] 90% of the 1,752 higher critical rated vulnerabilities that have been humanly reviewed were found to be accurate,” he said. “Cool, but that means it’s still about 15% of the total number that it found.”
According to Shipley, a critical question that has yet to be unanswered is the cost of finding each one of these vulnerabilities. “How many tokens are you burning? I’ve heard it’s in the range of $500 a minute, so I’m really curious to know what the cost is,” he said. “Surely, if they can tell us how ‘we found this many’, [they can answer] how much compute did it cost?”
He added that the only ultimate fix is to “make software makers liable for their software. That is the only way out of this mess, because that is the fundamental misalignment that got us here.”
View the full article
CSOonline

⚡ Weekly Recap: Linux Flaws, Defender 0-Days, Router Botnets, and Supply Chain Chaos

Monday recap. Same mess, new week. A sketchy dev tool got people pwned, old bugs came back from the dead, and security products somehow needed protecting from themselves. A bunch of companies spent the week checking old boxes and forgotten servers they should've patched years ago. Good times. Phishing crews are getting smarter too - less obvious scam junk, more targeted stuff that actuallyView the full article
Hacker News

Netherlands Seizes 800 Servers, Arrests 2 for Aiding Cyberattacks

Authorities in the Netherlands have arrested the co-owners of two related Internet hosting companies for operating IT infrastructure used by Russia to carry out cyberattacks, influence operations and disinformation campaigns inside the European Union. The two men were the focus of a 2025 KrebsOnSecurity story about how their hosting companies had assumed control over the technical infrastructure of Stark Industries Solutions, an Internet service provider sanctioned last year by the EU as a frequent staging ground for cyber mischief from Russia’s intelligence agencies.
An investigator with the Tax Intelligence and Investigation Service (FIOD), the Dutch financial crimes agency, during the raid. Image: FIOD.
The Dutch daily news outlet de Volkskrant reports that the Dutch financial crime agency FIOD on May 18 arrested a 57-year-old from Amsterdam and a 39-year-old from The Hague, charging them with violating sanctions law by directly or indirectly making economic resources available to EU-sanctioned entities.
The Dutch investigation focuses on Stark Industries, a sprawling hosting provider that materialized just two weeks before Russia invaded Ukraine. As detailed in this May 2024 deep-dive, Stark quickly became the source of massive distributed denial-of-service (DDoS) attacks against European targets, and emerged as a top supplier of proxy and anonymity services that showed up time and again in cyberattacks linked to Russia-backed hacking groups.
That report identified two Moldovan brothers — Ivan and Yuri Neculiti and their company PQHosting — who were providing one of Stark’s two main conduits to the larger Internet. In May 2025, the EU sanctioned PQHosting and the Neculiti brothers for aiding Russia’s hybrid warfare efforts. But as KrebsOnSecurity observed in September 2025, those sanctions failed to target Stark’s remaining connection to the Internet — an Internet service provider based in the Netherlands called MIRhosting.
MIRhosting is operated by Andrey Nesterenko, a 39-year-old Russian native who runs the business out of the Netherlands.  News that PQHosting and the Neculiti brothers were about to be sanctioned by the EU leaked in the media nearly two weeks before the sanctions were announced last year. During that time, the Stark network assets were transferred from PQHosting to a new entity called the[.]hosting, under the control of the Dutch entity WorkTitans BV.
And as our September 2025 report showed, WorkTitans was controlled by Nesterenko and a 57-year-old from Amsterdam named Youssef Zinad. On top of that, WorkTitans was getting connectivity to the larger Internet solely through MIRhosting, where Zinad had worked previously.
On May 18, Dutch financial crime investigators arrested Nesterenko and Zinad, and searched three businesses in Enschede and Almere and two data centers in Dronten and Schiphol-Rijk. A statement from the Dutch authorities said they also seized laptops, telephones and more than 800 servers.
A message to the-hosting customers immediately after 800 of its servers were seized by Dutch authorities. The message says that unfortunately data stored on the server has been lost and cannot be recovered.
De Volkskrant said it reviewed data showing WorkTitans and MIRhosting were the most-used networks in pro-Russian attacks on Danish government bodies between November 13 and 19, 2025, the week of Denmark’s municipal elections.
The publication wrote that prior to Nesterenko’s arrest, the MIRhosting founder denied that he knew his servers had been misused by pro-Russian cybercriminals. “He said he had ended all services with the Neculiti brothers when the EU sanctions came into force in May 2025,” and the he “reserved all rights to take action against ‘harmful and incorrect publications,” de Volkskrant wrote.
MIRhosting released a statement saying it has initiated an internal investigation into the alleged facts concerning the elections in Denmark, and that it has temporarily paused services to WorkTitans as a precautionary measure while the matter is being reviewed further.
“Based on our preliminary findings, there are no indications that the services over which we exercise control were actually used to influence the Danish elections,” the statement reads. “No anomalies or spikes were observed in our network traffic during the period mentioned in the publication; had large-scale DDoS attacks occurred, such activity would have been evident. Furthermore, prior to the media publication, we had not received any complaints, abuse reports, or official requests regarding suspicious activities or misuse of our network. Meanwhile, our regular operational activities continue, and our service to our other clients remains fully intact.”
Born in Nizhny Novgorod, Russia, Mr. Nesterenko grew up as a piano prodigy who performed publicly at a young age. In 2004, Nesterenko founded MIRhosting’s parent Innovation IT Solutions Corp., which has the notable distinction of being the company responsible for hosting stopgeorgia[.]ru, a hacktivist website for organizing cyberattacks against Georgia that appeared at the same time Russian forces invaded the former Soviet nation in 2008. That conflict was thought to be the first war ever fought in which a notable cyberattack and an actual military engagement happened simultaneously.
Responding to questions shared via email, Nesterenko said MIRhosting does not support cybercrime, sanctions evasion, or illegal activity, and that the allegations and arrest by Dutch authorities have been extremely harmful to him and his company.
“The transition to the.hosting was not intended to evade sanctions,” Nesterenko wrote. “The hardware and customer portfolio had already been transferred to WorkTitans before the sanctions appeared. Closing or damaging a legitimate Dutch infrastructure company will not stop cybercrime, but it will harm many people who have done nothing wrong.”
Far less is public about the 57-year-old Zinad, who reportedly has been keeping a low profile since our story last year. De Volkskrant reported that Zinad blocked access to his LinkedIn account, had gone months without responding to emails, WhatsApp messages and phone calls, and told a colleague that illness was forcing him to lead a somewhat more reclusive life.
Mr. Zinad’s now-defunct LinkedIn profile. It was full of posts for MIRhosting’s services.
Mr. Nesterenko claims Zinad was never an employee of MIRhosting.
“He helped me and MIRhosting with certain business tasks under a normal business-to-business arrangement between companies,” Nesterenko explained.
However, in previous emails to KrebsOnSecurity, Nesterenko carbon copied Mr. Zinad (who had a @mirhosting.com email), explaining that he was part of the company’s legal team. Also, the Dutch website stagemarkt[.]nl lists Youssef Zinad as an official contact for MIRhosting’s offices in Almere.
Mr. Zinad has never responded to requests for comment. Nor did de Volkskrant have any luck tracking him down. The publication said it repeatedly asked Mr. Zinad (referred to here as simply “Z”), but he reportedly avoided every form of contact.
“‘I am unavailable but will respond to your message as soon as possible,’ reads an automated reply on WhatsApp on 2 October 2025,” de Volkskrant reported. “It is the only response de Volkskrant would receive in months. He did not pick up his phone and did not call back. When an acquaintance asked him via LinkedIn to contact the reporter, he blocked access to his LinkedIn page. At an address in Almere where Z.’s personal limited company is registered, no one was present in April. The corner house’s blinds were drawn, and a pile of rubbish bags lay outside next to a container, as if someone had recently left. A neighbour said he knew the man but did not know where he was staying. Z. was later arrested at a residence in Amsterdam.”
View the full article
Krebs

AI security needs a shift from models to systems, researchers argue

Enterprises cannot secure AI agents by making the underlying models more robust and must instead enforce security controls at the system level around them, researchers behind a paper published this month argued, warning that traditional AI-security approaches are increasingly misaligned with how autonomous agents actually operate inside enterprise environments.
The paper argues that enterprises should stop treating AI agents as trusted software components and instead secure them as fundamentally untrusted systems operating inside enterprise infrastructure.
“The AI model powering the agent must be treated as an untrusted component,” the researchers wrote in the paper, warning that “semantic guardrails” and prompt-level defenses alone cannot reliably secure systems once agents gain access to enterprise tools, memory, APIs, browsers, and execution environments.
The authors drew the comparison to operating systems. “Similar to how an operating system treats a process as untrusted, we take the stance that the model powering the agent should be treated as untrusted and security properties should be expressed and enforced outside, at the level of the encompassing system,” they wrote.
The paper was written by researchers at Google, the University of California, San Diego, the University of Wisconsin-Madison, and other institutions, including Mihai Christodorescu, Earlence Fernandes, and Somesh Jha.
Five principles from systems security
The authors distilled five principles from decades of systems security research that they said agentic systems should follow: least privilege, tamper resistance of the trusted computing base, complete mediation, secure information flow, and accounting for the human as a weak link.
As evidence, the authors analyzed eleven real-world attacks on AI agents and mapped each to the principles it violated. The attacks included data exfiltration from the ChatGPT macOS app, a Claude Code exfiltration flaw, a Microsoft Copilot exfiltration vulnerability, and the AgentFlayer attack on Cursor through a malicious Jira ticket. Every one of the eleven violated the secure information flow principle, the paper said, while most violated the least privilege principle.
The authors rejected the idea that stacking machine-learning guardrails amounts to a defense.
“Merely stacking ML models does not constitute true defense-in-depth,” they wrote, because the guard models “often share the same statistical failure modes as the primary agents they monitor.”
To put the principles into practice, the authors proposed three security mechanisms, each tied to an open research problem the community has yet to solve.
The first is separating instructions from data, because language models mix the two in a single stream of tokens with no source-level distinction between them. The second is verifiable least-privilege policy generation, made difficult because security policies for agents are written in natural language and shift as a task evolves, which makes them hard to translate into rules a system can enforce. The third is information flow control, since tracking how sensitive data moves through a model remains unsolved.
Beyond the model
The paper challenges one of the dominant assumptions shaping enterprise AI-security efforts over the past two years — that increasingly capable models, alignment techniques, and prompt defenses would eventually make AI systems sufficiently secure for enterprise deployment.
Instead, the researchers argue AI agents should increasingly be treated more like operating environments or distributed systems than conventional enterprise applications because they combine reasoning, autonomy, memory persistence, and external tool execution inside a single operational layer.
“Security guarantees cannot emerge solely from better prompts, alignment tuning, or model-side mitigations,” the paper said, arguing enterprises instead need stronger runtime isolation, containment boundaries, least-privilege execution, and workflow observability controls around AI agents.
That creates situations where prompt injection is no longer simply a content-manipulation issue but potentially a workflow-execution and systems-integrity problem capable of influencing downstream actions across interconnected enterprise environments.
The visibility problem
The researchers also argue that current enterprise security tooling lacks sufficient runtime visibility into how AI agents actually reason, invoke tools, retain memory, and execute actions across enterprise systems.
Another paper published last week also points to a similar problem from a different angle, arguing that traditional endpoint detection and response platforms cannot adequately inspect AI-agent reasoning flows, prompt chains, memory interactions, or dynamic tool execution.
The paper proposed what researchers described as an “agentic detection and response or ADR” framework designed specifically for AI-agent environments.
“Current security tools are not designed to observe agent cognition or reasoning traces,” the researchers wrote, arguing that existing enterprise security stacks were built to monitor deterministic applications and endpoint activity — not systems capable of autonomous planning, probabilistic reasoning, and dynamic workflow orchestration.
The paper described a production deployment monitoring more than 10,000 AI-agent sessions daily across roughly 7,200 hosts, where researchers said the framework identified hundreds of credential-exposure incidents and other agent-related risks spanning 26 attack categories.
On a benchmark the team introduced, called ADR-Bench, the system detected 67% of attacks with zero false positives, outperforming three baselines, including Meta’s LlamaFirewall, by two to four times in F1-score, the paper said. On AgentDojo, a public prompt injection benchmark, it detected all attacks with three false alarms out of 93 tasks.
View the full article
CSOonline

Ghost CMS CVE-2026-26980 Exploited to Hijack 700+ Sites for ClickFix Attacks

Threat actors are exploiting a recently disclosed critical security flaw in Ghost CMS to inject malicious JavaScript code with an aim to fuel ClickFix attacks. According to QiAnXin XLab, the activity involves the exploitation of CVE-2026-26980 (CVSS score: 9.4), an SQL injection vulnerability in Ghost's Content API that could allow an unauthenticated attacker to read arbitrary data from theView the full article
Hacker News

As AI speeds coding, CVE Lite CLI keeps security deliberately AI-free

As AI coding assistants accelerate software development, one OWASP-backed open-source project is arguing that dependency security tooling still arrives too late to be truly useful.
CVE Lite CLI, a JavaScript and TypeScript dependency vulnerability scanner focused on local lockfile analysis, is positioning itself around a simple idea. Developers should see dependency risks while they are still writing code, not hours later inside a failing CI pipeline.
“What developers are missing is early feedback at the point where the dependency decision is made,” Sonu Kapoor, creator and maintainer of the project, told CSO. According to Kapoor, traditional CI-centric workflows often disconnect developers from the dependency choices that introduced risk in the first place.
CVE Lite CLI scans npm, pnpm, and Yarn lockfiles using OSV vulnerability data and claims to focus heavily on remediation guidance, including separating direct and transitive vulnerabilities, validating upgrade targets, and recommending actionable fix paths.
The project is being pitched as a “local-first” developer tool, as opposed to a replacement for enterprise software composition analysis (SCA) platforms, much like how developers already use ESLint or unit tests locally before CI runs them again later.
CVE Lite CLI targets an overlooked pain point
CVE Lite CLI is essentially trying to solve a workflow problem, Kapoor says many developers quietly struggle with. Dependency security checks often arrive after the work is already done.
The tool scans JavaScript and TypeScript lockfiles locally across npm, pnpm, and Yarn projects, so developers can understand dependency risk while they are still coding, not later in response to a failing CI pipeline.

Instead of just focusing on detection, the tool claims to look into subsequent questions like whether the issue is direct or transitive, whether there is a clean upgrade path, or whether upgrading one package actually removes the vulnerable dependency.
“In one real case, CVE Lite CLI skipped 27 package versions before finding a safer version to recommend,” Kapoor said, explaining the granularity of the tool. “That is the kind of work developers should not have to do manually by reading logs and retrying upgrades one by one.”
Kapoor said the tool can be configured for JSON, SARIF, or HTML outputs and can also be integrated into CI workflows as a GitHub Action.
AI could be making things worse
The argument arrives as software supply chain security continues to collide with AI-assisted development practices that allow developers to generate code, integrate packages, and restructure projects much faster than before.
Kapoor said this velocity changes the nature of dependency risk itself. “AI coding assistants have made this more important, not less,” he said. “That speed is useful, but it also means dependency decisions can happen quickly and sometimes without the same level of manual review. I do not think AI assistants remove the need for security checks.”
On the contrary, they increase the need for fast, local, explainable checks that can be run while the work is happening, he added.
One cited example involved scans against lint-staged, a widely used JavaScript tooling package. According to Kapoor, a standard “npm audit –omit=dev” workflow failed to surface a production dependency issue that CVE Lite CLI later identified through lockfile analysis. “Honestly, I don’t think most developers understand those blind spots in detail, and I do not mean that as criticism of developers,” he said. “The dependency graph in a modern JavaScript project is extremely noisy.”
A developer meaning to install one direct dependency may end up with hundreds or thousands of transitive packages.
CVE Lite CLI isn’t falling for AI
The project also deliberately avoids turning itself into a broader AppSec platform, despite growing industry pressure to consolidate security tooling into an AI-enabled ecosystem.
“I do think security tooling has become too heavy for the day-to-day developer workflow,” Kapoor said. “That does not mean those platforms are bad. It means they often serve security organizations better than they serve the individual developer trying to make a safe dependency decision during a normal coding session.”
This philosophy also extends to the project’s approach toward AI itself. While CVE Lite CLI includes integrations that help AI coding assistants interpret scan results, Kapoor said the underlying vulnerability analysis intentionally remains deterministic.
“I do not think AI should decide whether a CVE exists,” he said. “That part needs to be boring, repeatable, and auditable.”
Instead, the project uses AI as what the founder described as an “explanation and workflow layer” around scan results rather than as the scanner itself. “CVE Lite CLI includes AI assistant skills that teach tools like Claude Code, Codex CLI, Gemini CLI, Cursor, and GitHub Copilot how to run CVE Lite CLI, read its structured output, and help the developer understand or prioritize the remediation plan,” Kapoor explained.
Caution around expansion
Kapoor said he has been receiving positive feedback from the companies and developers using CVE Lite CLI in real workflows, asking him whether the same approach could support .NET or Python ecosystems.
“That interest is encouraging because it tells me the local-first, remediation-oriented model is resonating beyond the original JavaScript and TypeScript use case,” he said. “But I am cautious about expanding the current tool too broadly.”
The explanation he gave was simple. Each ecosystem, he believes, has its own package manager behavior, lockfile format, dependency graph semantics, advisory sources, and remediation patterns. “Adding those directly into CVE Lite CLI could make the tool heavier and less clear for the JavaScript and TypeScript developers it was originally designed to help.” The project has now been adopted into the OWASP foundation ecosystem as an official OWASP project and is available for free to developers on GitHub.
View the full article
CSOonline

Account

Navigation

Search

Search

Configure browser push notifications
Chrome (Android)
  1. Tap the lock icon next to the address bar.
  2. Tap Permissions → Notifications.
  3. Adjust your preference.
Chrome (Desktop)
  1. Click the padlock icon in the address bar.
  2. Select Site settings.
  3. Find Notifications and adjust your preference.