Today is all about who controls the future of AI, and at what price. Washington finally got its frontier-model order, but with a twist: it's voluntary, and labs get to decide whether Uncle Sam peeks under the hood. Meanwhile the money is moving in nine and twelve figures, Anthropic toward a near-trillion-dollar IPO, DeepSeek toward its first-ever raise, while Nvidia quietly puts the machine that powers all of it into full production. Oh, and Claude apparently spent the last few weeks finding ten thousand zero-days. Dig in.
Trump Signs AI Executive Order Creating a Voluntary 30-Day Frontier Model Review
On June 2, President Trump signed 'Promoting Advanced Artificial Intelligence Innovation and Security,' letting frontier developers voluntarily hand the federal government up to 30 days of pre-release access to probe new models for security weaknesses. Treasury, the NSA, and CISA are tasked with defining which models count as 'frontier.' It matters because it's the administration's first real federal AI oversight framework, and its voluntary design sidesteps the mandatory regime many expected, after Trump postponed the May 21 signing over concerns it could slow AI-driven job growth.
Anthropic Confidentially Files for IPO After Hitting a $965 Billion Valuation
Anthropic filed a confidential S-1 with the SEC on June 1, just four days after closing a $65 billion round that valued it at $965 billion, the highest valuation ever for a private AI company. Reports peg the listing window around October 2026, ahead of OpenAI's expected debut. The confidential draft lets the SEC review before the prospectus goes public; share count, price range, ticker, and exchange are all still undecided.
DeepSeek Pursues $7.4B in Its First-Ever Funding Round at Up to $59B Valuation
Chinese AI champion DeepSeek is in talks to raise roughly 50 billion yuan (about $7.4 billion) in its inaugural external funding round, targeting a post-money valuation of $52B to $59B. Founder Liang Wenfeng is committing 20 billion yuan of his own money, with Tencent (10B yuan) and battery giant CATL (5B yuan) lined up as the largest outside backers, plus talks with China's national AI fund, NetEase, and JD.com. The deal signals heavy domestic backing for the lab whose V3 and R1 models rattled Silicon Valley last year.
Anthropic's Claude Mythos Found 10,000+ Vulnerabilities, Expands Project Glasswing to 150 More Orgs
Anthropic revealed that its frontier 'Claude Mythos Preview' model used Project Glasswing partners to surface more than 10,000 high- or critical-severity software flaws, including thousands of previously unknown zero-days across every major OS and browser. Launch partners include AWS, Apple, Cisco, CrowdStrike, Google, JPMorganChase, Microsoft, and NVIDIA, and Anthropic is now extending the program to roughly 150 new organizations including critical-infrastructure operators. It matters because it reframes frontier AI as a defensive cybersecurity tool, while raising the obvious question of what the same capability does in the wrong hands.
Nvidia Puts Vera Rubin Into Full Production and Unveils Cosmos 3 at GTC Taipei
At GTC Taipei, Jensen Huang announced that Nvidia's Vera Rubin platform, the successor to Blackwell, has entered full production, with shipments slated for autumn 2026 and roughly 150 Taiwanese supply-chain firms powering the ramp. Vera Rubin pairs five purpose-built racks into one AI supercomputer delivering 10x the agent throughput of Grace Blackwell. Huang also debuted Cosmos 3, a frontier foundation model for physical AI that can act as a vision-language model, world model, simulator, or robot policy, with Super and Nano variants live on Hugging Face today.
