1. Trump labeled his own AI safety order an "innovation blocker" after the CEOs declined to attend the signing The signing event was on the calendar. An executive order mandating baseline safety testing for frontier AI systems was drafted.

2. Two-thirds of an AI chip's cost is now memory. Samsung's memory engineers just turned that into $340,000 bonuses A new analysis from Epoch AI puts memory at roughly two-thirds of the component cost of a modern AI accelerator. That share was much smaller when GPU compute first started scaling.

3. Microsoft revokes Claude Code from its own engineers as DeepSeek locks in its discount Microsoft started canceling Claude Code licenses for its own employees this month, The Verge reported, citing internal communications.

In Brief

• Google's AI answers swallow link clicks in search Google's AI-generated overviews now sit above traditional search results for most queries, cutting click-throughs to source sites. Wired argues users will adopt the feature regardless of preference because of its convenience, accelerating traffic loss for publishers and writers.

• xAI commits to natural gas, abandoning Musk's solar pitch xAI is powering data centers with natural gas turbines while SpaceX pursues orbital data centers. The choice contradicts the "solar-electric economy" Musk promoted before xAI's buildout.

• Reuters finds Grok barely used in US federal AI records Reuters reviewed federal AI usage logs from last year and found Grok almost absent, while ChatGPT, Claude, and Gemini dominate. xAI's chatbot trails competitors in both enterprise and consumer adoption despite Musk's continued promotion.

• OpenAI's Chris Lehane lobbies states against AI laws OpenAI global affairs chief Chris Lehane is working to soften state-level AI legislation that could constrain the company's growth. Lehane is also reframing public debate over AI's societal effects in OpenAI's favor.

• Hyperscalers press Gulf states to harden AI undersea cables Cloud and AI giants are pushing Gulf governments to rethink subsea cable infrastructure as AI workloads raise the cost of any outage. The region's hub status for new data centers depends on redundancy that does not yet exist.

• Spotify launches AI remix tool with Universal Music Spotify rolled out a tool that lets users generate AI remixes and covers of licensed songs, built with Universal Music Group. The company pitches it as a superfan feature, adding to AI-generated music already flooding YouTube, TikTok, and Instagram.

• Amazon's Bee wearable records audio continuously for AI Amazon's Bee wearable captures ambient audio to power AI assistant features such as summaries and reminders. TechCrunch's tester described useful outputs paired with concerns about always-on capture of conversations.

• Ferrari deploys IBM watsonx for F1 fan content Scuderia Ferrari HP is using IBM's AI to generate personalized race content and fan interactions during Grand Prix weekends. IBM and Ferrari briefed TechCrunch on how the system runs during live events.

• Author keeps using AI after book ships with fabricated quotes Steven Rosenbaum's book "The Future of Truth" contained AI-generated quotes that no real source ever said. Rosenbaum told Ars Technica he plans to continue using AI tools despite the published errors.

• San Francisco Tenderloin nonprofit replaces volunteers with kitchen robots A nonprofit in San Francisco's Tenderloin deployed robotic meal-prep equipment after volunteer numbers fell. The machines cook and assemble meals on site for residents the organization serves.

Read the full edition →