The Trump administration is asserting control over access to frontier AI models from companies like OpenAI and Anthropic, citing national security. This shift, formalized through the new "Gold Eagle" program, follows the temporary blocking of Anthropic's models and comes as Chinese AI like Moonshot AI's Kimi K3 becomes increasingly competitive, prompting warnings that the U.S. could lose its lead.
Government AI model gatekeeping
- ▪The Trump administration is now dictating which companies and entities are allowed access to new frontier AI models
- ▪Access to the Claude Mythos 5 and Fable 5 models was restored after weeks of negotiations between the Trump administration and Anthropic
- ▪A White House official stated that company engagements with government experts are "voluntary" and release decisions rest with the companies
- ▪In June 2026, the Trump administration blocked access to Anthropic's Claude Mythos 5 and Fable 5 models, citing "national security concerns."
- ▪Previously, AI companies like Anthropic and OpenAI controlled which entities had access to their most powerful models
Company-led security consortia
- ▪Future AI model rollouts will require explicit government approval regarding which partners can be included
- ▪OpenAI operates a similar consortium called Daybreak for its own cybersecurity model
- ▪Anthropic shared its Mythos cybersecurity model with select partners through a program called Project Glasswing
- ▪The future of company-led initiatives like Project Glasswing and Daybreak is now uncertain due to the administration's actions
Chinese AI competition
- ▪David Sacks, former White House AI czar, called the Kimi K3 breakthrough "concerning" and warned the US could lose the AI race
- ▪The Kimi K3 model's performance is comparable to Anthropic's Fable and OpenAI's GPT-5.6
- ▪On July 17, 2026, Chinese startup Moonshot AI unveiled its Kimi K3 model
- ▪The US government's actions come as cheaper, open-weight models from China are becoming competitive with American frontier labs
- ▪In at least one independent benchmark, Moonshot AI's Kimi K3 model outperformed U.S. frontier models
Gold Eagle clearinghouse program
- ▪The Gold Eagle program will function as a clearinghouse, giving the White House approval power over which companies can access new AI models
- ▪In July 2026, the Trump administration launched a program called "Gold Eagle" to collaborate with the private sector on cyber vulnerabilities
Frontier model release restrictions
- ▪The Trump administration asked OpenAI to gate access to its recent GPT-5.6 model release
- ▪In June 2026, OpenAI announced it would limit new AI models to "trusted partners" to comply with government requests
Story comments
Loading comments…