The Omniscient Bulletin · 2026-06-26
The Omniscient Bulletin — June 26, 2026
Washington holds back OpenAI's next model as California starts counting AI's hit to jobs
On Thursday the government stepped into the AI story on two fronts. Washington asked OpenAI to hold back its next model, GPT-5.6, until officials can vet its cyber capabilities, the first time the US has restricted an American model before release. The same day, California became the first state to measure AI's hit to jobs. Neither slowed the market: over a billion dollars flowed into companies built to automate human work, from a Bezos- and Schmidt-backed lab to a startup vowing to handle a billion customer-service calls. Underneath, IBM claimed the first sub-1-nanometer chip and Amazon pledged $13 billion more for Indian data centers.
Policy
Trump administration asks OpenAI to stagger the release of GPT-5.6
The Trump administration asked OpenAI to limit its next model, GPT-5.6, to a small set of government-approved partners before any wider release, the first such US pre-release restriction. The White House cyber and science-policy offices made the request while building a model-testing protocol; a source said GPT-5.6 has 'Mythos-like' capability. In a memo reported by The Information, Sam Altman called it 'not our preferred long term model' and hopes to widen the launch a couple of weeks later. It follows the June 2 executive order on voluntary pre-release testing.
Labor
California becomes the first state to track AI's hit to jobs
California launched what it calls the nation's first state tool to track AI-driven job loss, a public dashboard from the California Policy Lab at UCLA and the state Employment Development Department. It ties the occupations most exposed to AI to monthly unemployment-claims data as an early warning system. The first read is measured: no statewide surge in AI layoffs, but elevated claims among college-educated workers in high-exposure roles and among Bay Area tech workers since ChatGPT arrived in late 2022. It implements an executive order Governor Newsom signed in May.
Office of the Governor of California→
Industry
General Intuition raises $320M with Bezos and Schmidt to build action models
General Intuition disclosed a $320 million round at a $2.3 billion valuation, led by Khosla Ventures with Jeff Bezos, Eric Schmidt, and former F1 champion Nico Rosberg among the backers. The 25-person startup, a sister company to the gaming-clip platform Medal, trains what it calls large action foundation models on billions of gameplay clips, betting that learning to act in virtual worlds transfers to robots and agents. The round closed in January and was disclosed this week, lifting its total raised to about $454 million.
Industry
Scaled Cognition raises $100M for enterprise AI agents built not to hallucinate
Scaled Cognition raised a $100 million Series A led by Khosla Ventures for APT, an Agentic Pretrained Transformer it pitches as a model that will not give a wrong answer. The company says customers deploying it are on track to automate more than a billion customer-service interactions within a year, targeting the $600 billion business-process-outsourcing market. It was cofounded by Berkeley AI professor Dan Klein and Dan Roth, who earlier sold an agentic AI startup to Microsoft. Genesys is both a customer and an investor.
Scaled Cognition / GlobeNewswire→
Industry
France's Alan raises 480M euros at a 5.5B euro valuation for AI health insurance
Alan, the French digital health-insurance startup, raised 480 million euros in a Series G valuing it at 5.5 billion euros. Prosus led the round with a 400 million euro primary investment, joined by Index Ventures, Teachers' Venture Growth, and new backer Dara Holdings. Alan reported more than 800 million euros in annual recurring revenue as of Q1 2026, up 53% year-on-year, serving over 1.1 million members across France, Belgium, Spain, and Canada. The funds will support international expansion and AI product development.
Compute
IBM unveils what it calls the world's first sub-1nm chip technology
IBM said it has built the world's first sub-1-nanometer chip technology, a 0.7-nanometer node using a new 3D design it calls nanostack that vertically stacks and staggers transistors to pack nearly 100 billion onto a fingernail-sized chip. That is about double the density of its 2021 two-nanometer chip and, IBM says, yields up to 50 percent more performance or 70 percent better energy efficiency. Developed at its Albany, New York labs with ASML, Lam Research, and Tokyo Electron, the design has been validated in the lab, with production as early as five years out.
Compute
Amazon pledges another $13B for AI data centers in India
Amazon said it will invest a further $13 billion through 2030 to expand AWS data centers in Mumbai and Hyderabad, announced after chief executive Andy Jassy met Prime Minister Narendra Modi in New Delhi. The commitment lifts Amazon's planned India spending to about $48 billion through the end of the decade, underlining how hard the big cloud providers are racing to plant AI capacity there. Microsoft and Google have made their own multibillion-dollar India pledges, betting the country becomes a major hub for AI computing.