The Trump administration is asserting greater oversight of access to cutting-edge artificial intelligence models, shifting decision-making away from leading developers, according to people familiar with the matter. For years, companies such as OpenAI and Anthropic have decided which partners could use their most advanced systems through security-focused programs. While the White House says company participation remains voluntary under a June executive order, officials last month temporarily blocked releases of Anthropic’s Claude Mythos 5 and Fable 5 on national-security grounds, with access restored after negotiations. OpenAI has since limited distribution of new models to “trusted partners” to align with government requests.
The administration this week launched “Gold Eagle,” an AI clearinghouse intended to coordinate with the private sector on cybersecurity testing. People familiar say the program could put the government in a position to greenlight which firms receive early access to new models, raising questions about the future of company-led initiatives like Project Glasswing and Daybreak.
The moves arrive as Chinese competitors narrow the performance gap. Moonshot AI’s Kimi K3 model, unveiled Friday, matched or surpassed U.S. frontier systems on some benchmarks, drawing concern from industry figures who warn that added oversight could slow domestic innovation. The White House says it aims to strengthen security without stifling progress, but enterprises may face new uncertainty around timelines and eligibility for access to next-generation AI.
Trump administration moves to control who gets access to frontier AI models, sources say
- Trending
- Comments
- Latest
