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

• The White House blocked Anthropic's Claude Mythos 5 and Fable 5 releases last month due to 'national security concerns' and required OpenAI to limit new AI models to U.S. partners only
• The administration's new 'Gold Eagle' clearinghouse program will control which companies receive access to new AI models, putting company-led initiatives like Project Glasswing and Daybreak in doubt
• Chinese startup Moonshot AI's Kimi model matched or outperformed GPT-5.6 and Fable in benchmarks, with former White House AI czar calling the development 'concerning' for America's AI leadership

Summary: White House Asserts Control Over Frontier AI Model Access

The Trump administration is seizing greater control over the release and distribution of advanced AI models, shifting decision-making power away from technology companies, according to CNBC sources. Previously, companies like Anthropic and OpenAI independently determined which organizations could access their most powerful models through initiatives like Project Glasswing and Daybreak.

Key Developments:

The White House this week launched "Gold Eagle," a clearinghouse program that positions the government to approve which companies can access new frontier AI models. This follows last month's blocking of Anthropic's Claude Mythos 5 and Fable 5 releases due to "national security concerns" after extensive negotiations. OpenAI has also limited new model releases to comply with government requests.

Official Position:

A White House official stated that government engagement is "voluntary" and release decisions rest with companies, referencing President Trump's executive order from June requesting voluntary early government access for testing. The administration claims it doesn't provide formal approvals for private company AI releases.

Market Implications:

The policy shift comes amid heightened cybersecurity concerns and intensifying global AI competition. Chinese startup Moonshot AI released its Kimi model Friday, matching or exceeding performance of U.S. frontier models in some benchmarks. Former White House AI czar David Sacks criticized the approach, warning "This is how you lose the AI race."

Industry Impact:

The future of company-led initiatives like Project Glasswing and Daybreak remains uncertain, with sources indicating rollouts will require explicit government approval for partner involvement. This represents a significant power shift from tech giants to federal oversight in the rapidly evolving AI sector.