/https://media.cnbc.com/dims/origin/108343257/Spider-Man-Brand-New-Day.jpg "Spider-Man Smashes Box Office Records with ‘Brand New Day’ Preview Haul")
The Trump administration is taking unprecedented steps to control access to cutting-edge artificial intelligence models, effectively shifting power from leading tech firms like Anthropic and OpenAI. Citing national security concerns and intensifying global competition, particularly from China, the White House is implementing new oversight initiatives like “Gold Eagle” that will dictate which partners can access new AI releases, raising questions about industry-led programs.
In a significant shift of power, the Trump administration is increasingly dictating who gets access to the most advanced frontier artificial intelligence models, wresting control that was previously held by leading American AI companies.
Historically, tech giants like Anthropic and OpenAI had autonomous decision-making power over the distribution of their most powerful AI models, often through exclusive initiatives such as Anthropic's Project Glasswing for its Mythos cybersecurity model and OpenAI's Daybreak consortium for its own cybersecurity AI.
However, CNBC has learned that the White House is now asserting greater authority. While a White House official stated that engagements with government experts are "voluntary" and that "decisions on timing and scope of releases rest entirely with the companies," the reality on the ground suggests a different dynamic. Last month, the administration temporarily blocked Anthropic's Claude Mythos 5 and Fable 5 due to "national security concerns," only reinstating access after intense negotiations. Similarly, OpenAI recently announced it would limit new AI models to "trusted partners" following government requests.
This increased oversight comes at a critical juncture, as sophisticated AI tools present growing cybersecurity risks and global competition intensifies. Chinese startup Moonshot AI recently unveiled its Kimi K3 model, which has largely caught up to, and in some benchmarks even surpassed, the performance of leading U.S. frontier models like Fable and GPT-5.6.
David Sacks, founder of Craft Ventures and former White House AI czar, voiced his concern, stating that this is "how you lose the AI race" if the U.S. bogs itself down while other nations don't play by the same rules.
The administration's strategic moves include President Donald Trump's June executive order, which requested voluntary early access to models for government testing. More recently, the White House launched "Gold Eagle," a program designed to collaborate with the private sector to identify and rectify cyber vulnerabilities in AI. Sources familiar with the matter indicate that this new clearinghouse will put the White House in charge of greenlighting which companies can access novel AI models.
The future of previously company-controlled initiatives like Project Glasswing and Daybreak now appears uncertain, with new rollouts reportedly requiring explicit government approval for partner involvement. This marks a clear pivot towards greater government intervention in the burgeoning field of artificial intelligence.
VIDEO: Trump signs AI executive order asking companies to give government early access to models
(Available at CNBC, duration 1:49. Thumbnail also serves as the main image above.)
