The White House is now dictating access to frontier AI models, shifting power away from tech giants

The Trump administration has moved to assert direct control over which organizations get access to new frontier AI models, according to CNBC, launching a program called Gold Eagle this week that functions as a clearinghouse for AI related cyber vulnerabilities and, according to a source familiar with the program, will put the White House in charge of greenlighting which companies can access newly released AI models. Until now, that decision sat entirely with the labs themselves: Anthropic and OpenAI have each decided independently which enterprise customers, government agencies, and other organizations get early or privileged access to their most capable models. A White House official pushed back on the framing when asked by CNBC, saying the administration does not provide approvals for AI releases from private companies. The shift comes against a backdrop of intensifying competition with China, days after Moonshot AI's Kimi K3 largely matched the performance of Fable 5 and GPT 5.6 and even outperformed both on at least one independent benchmark, a result that has sharpened Washington's sense that frontier AI capability is now a contested strategic asset rather than a purely commercial one. If Gold Eagle functions as described, it would mark one of the most direct interventions yet by the federal government into how private AI labs decide who gets to use their own most powerful models.