Microsoft is adding per-process neural processing unit monitoring to Windows Task Manager, giving users a way to view AI-related hardware usage at the process level. The feature extends the long-running Windows utility to surface NPU workload activity alongside other system resource metrics.
TrueFoundry has released TrueForge, an open-source AI agent harness under the MIT License, aimed at giving developers and enterprises more control over agent behavior, tooling, and hosting. The company says benchmark tests on DevRev's Enterprise-Bench showed lower task-completion costs than Anthropic's Claude Managed Agents, including reported savings when using both GLM-5.2 and Claude Opus 4.8. The article centers on a discrete product release and performance claim from TrueFoundry.
TechCrunch reports that Cognition CEO Scott Wu has denied a report claiming SpaceX tried to acquire the AI coding startup. The article describes the denial as a discrete public response to acquisition speculation involving Cognition and SpaceX.
The Wall Street Journal reports that flight attendants at bankrupt Spirit Airlines are seeking assurances that confidential personal information will be excluded from any sale of the airline’s digital records to Google for AI-related use. The dispute centers on privacy and data-handling terms tied to a reported bid for access to Spirit’s data assets.
Futurism reports that the European Central Bank warned that a sharp reversal in AI-related market enthusiasm could have wider financial spillover effects beyond the United States. The article frames the warning as a discrete institutional risk signal from a major central bank rather than a new AI product, funding round, or company action.
The Register reports that Google is considering Marvell as an alternative or counterweight to Broadcom in its AI chip supply strategy. The report says Marvell has proposed a $12.2 billion stake transaction to strengthen its position with Google as demand for AI infrastructure and custom silicon continues to rise.
Google has launched new study-focused AI features across Google Search and Gemini, targeting students and learning use cases. The update expands Google's effort to position Gemini as a consumer learning assistant while integrating more AI help directly into its core search product.
Google is offering eligible college students one year of its AI Pro subscription at no cost, according to Engadget. The promotion waives a plan that would normally cost about $200 and appears aimed at increasing student adoption of Google's paid consumer AI tools.
TechCrunch reports that some researchers said OpenAI revoked their access to the company’s limited Trusted Access for Cyber program, which is intended to give selected defenders stronger models for vulnerability research and responsible disclosure. The report frames the change as a dispute over access management within a controlled cybersecurity initiative rather than a new model launch.
Meta has launched a standalone Meta AI app for Mac, giving its AI assistant a native macOS client. The app supports sharing a user’s window with Meta AI so it can answer questions, suggest actions, or generate content based on on-screen context, and it also adds dictation support across apps.
Google is participating in Operation Blue Skies, a project involving the UK government and airline partners that will use AI-informed analysis to slightly reroute flights in an effort to avoid forming contrails. The tests focus on whether targeted path changes can reduce aviation warming effects without major operational disruption.
Wired reports that developers began sharing workarounds for Anthropic’s invisible watermarking of Claude-generated text within hours of the feature being announced. The article frames this as an early real-world test of the company’s watermarking approach, which Anthropic said it introduced to help meet European AI transparency requirements.
Wired reports that Flock Safety has developed a more advanced police AI investigation system that goes beyond automatic license plate recognition and is already being used by some law enforcement agencies. The outlet says it reconstructed the tool's code to examine how the product works and to assess the scope of its surveillance and investigative capabilities.
Z.ai has made its GLM-5.3 model available through an API, giving developers programmatic access to the newly released model for agents and applications. According to the report, pricing remains unchanged from GLM-5.2 at $1.40 per million input tokens and $4.40 per million output tokens, with lower cached-input pricing and free cached-input storage listed for a limited time.
Cerebras has introduced CS-4 rack systems that, according to The Register, double per-chip performance and fit roughly three times as many chips into a rack as the prior generation. The announcement points to a new hardware generation aimed at improving AI compute density and extracting more performance from Cerebras' wafer-scale architecture.
The Register reports that OpenAI is expanding multistage monitoring around chain-of-thought handling for frontier-model work, a change that raises operating overhead by roughly 20 percent for some workloads. The reported move appears to be an operational security and safety measure rather than a new model or product launch.
Block has open-sourced Berd, a desktop application it originally built for internal use as a unified workspace for AI agents, models, files, and projects. The app is available on GitHub under the Apache 2.0 license with downloadable builds for macOS, Windows, and Linux, and Block says it stores conversation history locally while working directly with local files, repositories, and tools.
Robin Williams' children, Zak, Zelda, and Cody Williams, have taken over the late actor's Instagram account and said they want it to serve as a safe, trusted place amid concerns about AI misuse of his likeness. The move follows public objections from Zelda Williams to unauthorized or unsettling uses of AI-generated representations of her father.
Harvey has updated its AI legal platform to remember individual users' preferences and work context, according to The Wall Street Journal. The report says the company is using memory to personalize legal workflows, with broader organizational memory described as a longer-term goal.
OpenAI has introduced a version of ChatGPT tailored for teenagers, according to The Wall Street Journal. The experience reportedly adds student-oriented features such as homework help and break reminders, alongside stronger guardrails intended to reduce emotional dependence and make the product more age-appropriate.
