The Trump administration has asserted control over access to frontier AI models from American companies, two sources familiar with the matter told CNBC. Previously, Anthropic and OpenAI independently decided which companies and agencies could access their most powerful models. Last month, the administration blocked Anthropic's Claude Mythos 5 and Fable 5 due to national security concerns, reinstating access after weeks of negotiations. The shift marks a fundamental change in oversight as sophisticated AI tools pose cybersecurity risks while Chinese competitors rapidly advance.
Trump Administration Blocks and Reinstates Anthropic Models
Last month, the Trump administration blocked Claude Mythos 5 and Fable 5 due to "national security concerns," reinstating access after weeks of intense negotiations with Anthropic. Anthropic unveiled its Mythos cybersecurity model to a handful of partners with Project Glasswing. A White House official told CNBC that the administration doesn't provide approvals for AI releases from private companies, stating any engagements are "voluntary" and "decisions on timing and scope of releases rest entirely with the companies."
OpenAI Limits Model Access Following Government Request
OpenAI last month said it would limit new AI models to "trusted partners" to comply with government requests. The company was asked by the administration to gate its recent GPT-5.6 release. OpenAI has a consortium called Daybreak for its cybersecurity model. The administration's moves have left the future of company-led initiatives like Project Glasswing and OpenAI's Daybreak in doubt.
Chinese Moonshot AI Unveils Kimi K3 Model
Chinese startup Moonshot AI unveiled its Kimi K3 model on Friday, which largely caught up to the performance of Fable and GPT-5.6, and even outperformed the U.S. frontier models in at least one independent benchmark. David Sacks, founder of Craft Ventures and the former White House AI czar, called the Kimi breakthrough "concerning." "This is how you lose the AI race," he wrote. "The rest of the world won't play by our rules if we bog ourselves down."
White House Launches Gold Eagle Cybersecurity Program
This week, the administration launched its own program, dubbed "Gold Eagle," aimed at collaborating with the private sector to find and fix cyber vulnerabilities. The clearinghouse would put the White House in charge of greenlighting which companies can access new AI models, according to a source familiar with the matter. President Donald Trump issued an executive order in June asking companies to voluntarily give the government early access to models for testing. Going forward, according to one source, these rollouts will require explicit government approval for which partners can be involved.
FAQ
What did the Trump administration do with Anthropic's AI models last month?
Last month, the Trump administration blocked Anthropic's Claude Mythos 5 and Fable 5 due to national security concerns, reinstating access after weeks of intense negotiations with the company.
Why did OpenAI limit access to new AI models?
OpenAI last month said it would limit new AI models to "trusted partners" to comply with government requests. The administration asked OpenAI to gate its recent GPT-5.6 release.
What is the Gold Eagle program launched this week?
This week, the administration launched Gold Eagle, a program aimed at collaborating with the private sector to find and fix cyber vulnerabilities, with the White House in charge of greenlighting which companies can access new AI models.
