First Washington Decided Who Could Run AI. Now OpenAI Wants to Give It a $42 Billion Ownership Stake — as a Gift.
A few weeks ago the US government stopped regulating AI and started allocating it — deciding who could run which model. Now OpenAI proposes handing the government an ownership stake in the industry: 1-5% of itself, up to ~$42.6B, donated (not sold), to seed a 'Public Wealth Fund,' framed as a template for every leading lab. Read past the civic branding and it's the regulated gifting the regulator a stake — reverse regulatory capture, at the moment the state controls frontier access and OpenAI is about to IPO. The governance conflict (a regulator that profits from the labs it gates) is acute, and who votes the shares — OpenAI's upside-without-control vs Sanders' 50%-with-control — is the whole fight.
The Government's Big Move to Solve AI's Power Crunch Speeds Up the Paperwork. It Doesn't Add a Single Watt.
FERC unanimously ordered all six US grid operators to fast-track how they connect large AI data centers — a mandate reaching two-thirds of US electricity demand and 200 million people. It sounds like the cavalry. The fine print: it fast-tracks the connection *process*, not the power *supply*. FERC can compress a multi-year interconnection queue and reassign who pays for the wires, but it cannot add one watt of generation to a grid that's short. That's why the hyperscalers already routed around it — building their own gas now and betting on nuclear by 2030. Washington sped up the line to a well that's running low.
The World's Big AI-Safety Report Said Autonomous Cyberattacks Aren't Here 'Yet.' The 'Yet' Is Expiring Faster Than the Report Can Be Updated.
In February, 100+ experts from 30+ countries, chaired by Yoshua Bengio, published the closest thing to a scientific consensus on AI risk. Its careful cyber judgment: AI already automates *most* of an attack and runs *some* steps autonomously, but no fully end-to-end autonomous attack has been reported — 'not yet.' Five months on, that 'yet' is under visible strain from developments the report predates (autonomous zero-day models, a real government breach). The real story isn't the cyber finding; it's that you can't govern a months-fast capability curve with a once-a-year consensus document. Also: contrary to the framing going around, China participated — Russia is the absentee.
The EU's Big AI-Transparency Deadline Hits August 2. The Part That Actually Matters Quietly Slipped to December.
On August 2 the EU AI Act's Article 50 transparency rules take effect — and what actually arrives is a chatbot telling you it's a chatbot. The version in the headlines, sweeping machine-readable labeling of AI-generated content, is not what goes live: that provision (50(2)) was quietly deferred to December 2 for every genAI system already on the market. Live Aug 2 is cheap disclosure (chatbot ID, deepfake disclosure, emotion-recognition notice); the hard provenance piece — the one that would let anyone tell real from synthetic — slipped four months, and even then nobody's solved how to label a half-AI-edited video.
The AI-Jobs Study Everyone Quoted Says High Earners Get Hit First. It Mostly Measured Who Uses Claude.
Anthropic's labor-market study — the most-cited AI-and-jobs evidence of the year — found the most-exposed workers are educated, higher-paid professionals, programmers first, inverting the low-wage-first narrative. Genuinely interesting. But its empirical core is Claude usage among Anthropic's own users, not the economy, so 'observed exposure' is partly a map of who uses Claude. There's no unemployment spike, so 'Great Recession for white-collar workers' was a headline, not a finding. The real number is the 94%-vs-33% gap between what AI can do and what it's actually doing — and the conflict of a lab grading its own product.
The Statute That Pulled Claude Offline Was Written for Physical Goods. Nobody Is Sure It Applies Here.
On June 12, Commerce Secretary Howard Lutnick signed a Bureau of Industry and Security directive forcing Anthropic to suspend Claude Fable 5 and Mythos 5 globally. Both models remain offline on day 13. The directive cited EAR § 744.22(b) — a statute written to control physical exports and hardware — and applied it to API inference access to a cloud-hosted language model. Whether that application is legally valid is an open question. If it is, BIS now has authority to take any cloud AI service offline for any foreign national, with no advance notice and no appeal mechanism in the statute. That is a bigger story than any single model suspension.
FERC Built a Fast Lane for AI Data Centers. It Leads to a Grid That Doesn't Have Enough Power.
FERC unanimously ordered all six major US grid operators to fast-track AI data center interconnection on June 18 — a significant regulatory action using emergency authority that bypasses the standard 2-5 year rulemaking process. FERC's own order confirmed it 'did not address the shortage of generating capacity.' The fast lane is real. The bottleneck it leads to is also real. And the renewable energy projects that would relieve that bottleneck have been dying in the same queue that AI data centers just jumped.
America Has Two AI Governance Regimes. They Contradicted Each Other Within Ten Days.
On June 2, the Trump White House signed an executive order creating a voluntary AI review framework and explicitly barring mandatory licensing. Anthropic called it 'an important step.' Ten days later, the Commerce Department suspended Anthropic's two flagship models globally under a separate national security authority — no voluntary framework, no 30-day window, no appeal process. The EO and the export control directive came from the same administration. They are not the same policy.
EU Proposes First Agent-Specific Regulations Under AI Act Framework
The European Commission has unveiled draft regulations specifically targeting autonomous AI agents, marking the first regulatory framework to distinguish between single-turn AI systems and multi-step agentic workflows. The proposal includes mandatory risk assessments, human oversight requirements, and liability rules for agent deployments in high-stakes domains.