US President Donald Trump points his finger as he signs an executive order on AI next to US Senate Commerce Committee Chairman Ted Cruz (R-TX) and US Commerce Secretary Howard Lutnick, in the Oval Office at the White House in Washington, DC, US Dec. 11, 2025.
Al Drago Reuters
President Donald Trump on Tuesday signed an executive order asking intelligence companies to provide models to the federal government to test their capabilities before full release.
The order asks companies, on a voluntary basis, to participate in an evaluation process to assess “enhanced Internet capabilities” and to determine whether it should be considered a “covered border model.” It then requests access to those models up to 30 days before companies plan to release them more widely, and enables the government to help select “trusted partners” who will get early access.
“Nothing in this section shall be construed as authorizing the creation of mandatory government licensing, pre-specification, or approval requirements for the development, publication, release, or distribution of new AI models, including frontier models,” the order said.
Trump signed the order in private, weeks after canceling a signing ceremony with senior technology officials because he “didn’t like certain aspects,” he told reporters at the time.
The order, which is thin on specifics, comes at an important time for AI development in the US
On Monday, Claude developer Anthropic said it had filed privately with the Securities and Exchange Commission for an IPO, while rival OpenAI is also preparing for a possible offering this year. Elon Musk’s SpaceX, owner of the AI lab xAI, is ready to hit both of them on the public market, with a debut as soon as next week that could value the company at more than $1 trillion.
The tech industry, whose fortunes have soared during the AI boom, has played a key role in the White House’s positions on AI. Venture capitalist David Sacks, a long-time partner of Musk’s, served as the first chief of crypto and AI before that role ended earlier this year. But sacks, and Musk once Meta CEO Mark Zuckerberg reportedly called the Trump administration last month to persuade them to oppose the AI executive order the president was set to sign.
The order also comes after Anthropic took government officials and Wall Street to task earlier this year by announcing the Claude Mythos Preview, a model that excels at identifying vulnerabilities and security flaws within software. The company limited the release to a select group of companies as part of a cybersecurity program called Project Glasswing, which it expanded on Tuesday.
The Mythos launch prompted several high-profile meetings between Anthropic and senior members of the Trump administration, including Susie Wiles, White House chief of staff, and US Treasury Secretary Scott Bessent.
WATCH: Trump signs AI asking companies to give government early access to models
