1. Two months after Trump called it "Leftwing nut jobs," Anthropic hands Washington a red-team model For nearly two months, the Trump administration has publicly attacked Anthropic.

2. Chrome tabs and Google Photos are now Gemini's default context Google rolled out two updates this week that wire personal data directly into its AI stack.

3. Stare into the orb, get five free Tinder boosts A Tinder user in Japan or Argentina can now walk up to a chrome sphere called the Orb, let it scan their iris, and walk away with five free Boosts — the in-app perk that pushes their profile to the

In Brief

• Cursor seeks $2B round at $50B valuation Cursor is in talks with a16z and Thrive to raise over $2 billion at a $50 billion valuation. Returning backers are expected to lead as enterprise revenue accelerates.

• OpenAI shuts Sora, loses Kevin Weil and Bill Peebles Kevin Weil and Bill Peebles are departing OpenAI as the company closes Sora and folds its science team. Leadership is redirecting resources away from consumer video toward enterprise AI.

• Construction delays hit 40% of planned 2026 US data centers Satellite and drone imagery show roughly 40% of data centers slated for 2026 completion are falling behind schedule. Power grid capacity and local permitting resistance are the main bottlenecks.

• Meta's AI data center push raises Quest headset prices Component costs for Meta's Quest headsets are climbing because the same memory and silicon feed its AI buildout. Consumers will pay more for VR hardware as a direct result.

• Analysis: "tokenmaxxing" coders ship more code but spend longer fixing it A TechCrunch analysis argues developers using high-token AI workflows generate more code while spending additional time rewriting and debugging it. Raw output volume masks falling net productivity.

• Mozilla ships Thunderbolt self-hosted AI client Mozilla released Thunderbolt, an AI client built on deepset's Haystack framework aimed at self-hosted deployments. The project pitches itself as an open alternative to centralized assistant services.

• Google blocked 8.3B ads in 2025 while suspending fewer advertisers Google blocked 8.3 billion ads last year but banned fewer advertiser accounts than the prior year. The company says AI enforcement now catches bad ads rather than bad actors.

• Microsoft and Stellantis embed AI across Jeep and Peugeot brands Microsoft and Stellantis announced an AI partnership covering digital services for brands from Jeep to Peugeot. Owners will see AI-assisted features in navigation, maintenance, and vehicle support.

• Luma opens AI production studio, first film stars Ben Kingsley as Moses Luma launched an AI-powered studio called Wonder Project. Its debut feature about Moses stars Ben Kingsley and releases this spring on Prime Video.

• Dairy Queen deploys Presto AI chatbot across drive-thrus Dairy Queen is rolling out AI chatbots to dozens of US and Canadian drive-thrus using Presto. The system takes orders and prompts customers to add more items.

• Anthropic publishes Claude design system Anthropic released Claude Design, documenting the visual identity and design principles behind its assistant. The post describes typography, color, and motion choices made by Anthropic Labs.

• ByteDance's Seedance 2.0 paper unifies text, image, audio, and video inputs ByteDance researchers published Seedance 2.0, a multi-modal generation model accepting text, image, audio, and video inputs. The Chinese release targets joint audio-video output at production scale.

Read the full edition →