Trump AI executive order faces deadline as AI regulation fight grows

The Trump AI executive order reaches a 60-day implementation deadline on Saturday, with federal agencies due to produce a framework for reviewing advanced artificial intelligence models. The outcome could affect how OpenAI, Anthropic and other developers release frontier systems, and how U.S. officials treat fast-improving open-weight models from China.

President Donald Trump signed the order in early June. The White House order asks AI companies to submit models voluntarily for government evaluation before public release and directs senior officials to define the process.

Those officials include Treasury Secretary Scott Bessent, Defense Secretary Pete Hegseth, White House Chief of Staff Susie Wiles and Commerce Secretary Howard Lutnick, according to the order. The White House has not publicly released the completed framework.

Asked by CNBC about the status of the work, the White House pointed to a social media post by spokesperson Kush Desai, who wrote: “BREAKING: Trump White House to meet a deadline we set for ourselves.” CNBC reported that representatives for OpenAI and Anthropic did not immediately respond to requests for comment.

What is in Trump's AI executive order?

The executive order calls for a classified benchmarking process to assess advanced models’ cyber capabilities and determine whether they qualify as a “covered frontier model.” It also says companies should work with the government on identifying “trusted partners” that may receive access to those models.

In practice, such a framework would set the government’s process for reviewing powerful AI systems before or during deployment. The order describes the company submissions as voluntary, but recent rollouts show that requests from U.S. agencies can still shape how quickly models reach customers.

After the order was signed, Anthropic disabled access to its Fable 5 and Mythos 5 models to comply with a Commerce Department export-control directive, CNBC reported. OpenAI later said it would limit the rollout of its GPT-5.6 model series to “trusted partners” at the request of the U.S. government. CNBC reported that both companies later expanded access more broadly.

OpenAI Chief Executive Sam Altman told CNBC on Wednesday that he had reviewed a draft of the framework and did not have “anything specifically in mind” for changes. Altman met with Wiles during a Washington visit and was expected to meet Bessent and Lutnick, according to reports cited by CNBC.

Why are open-weight AI models part of the debate?

Open-weight models allow users to download, modify and run AI systems on infrastructure they choose. Models largely developed in China have advanced quickly and offered lower-cost alternatives to closed frontier systems from companies such as OpenAI and Anthropic, intensifying pressure on U.S. policymakers.

The possibility that the Trump administration could restrict or ban Chinese open-weight models has drawn opposition from several major technology companies. Nvidia Chief Executive Jensen Huang published a letter on Monday urging policymakers to avoid “premature restrictions” on open-weight models.

Elon Musk responded on X that Huang had his “full support” and that “Jensen is right.” Microsoft, Meta, Palantir and other companies signed the letter. OpenAI later added its name, while Anthropic set out its position separately in a blog post, according to CNBC.

The administration has not announced an official position on a potential ban. CNBC reported that David Sacks, the former AI czar who remains influential in the administration, has been among the most vocal opponents of such restrictions.

Wiles wrote on X in late June that the government and private sector had worked together in an unprecedented way and that their shared aim was to deploy “the best tech” quickly and safely. The framework due Saturday will show how that objective is translated into rules for model reviews, partner access and national-security scrutiny.

This story draws on original reporting from CNBC.