The Trump administration is considering a new order that would allow US security agencies to vet powerful artificial intelligence models before companies release them to the public.
The plan was rolled out at a White House press conference led by the Office of the National Cyber Director. The meeting included OpenAI, Anthropic, and Reflection AI, all of which are private companies, so there are no stock tickers for them.
Donald Trump could sign the order on Thursday. After the first mention, Trump is the name used here. The plan will establish a “voluntary framework” for companies building leading-edge AI systems.
Under this setup, AI companies would notify the US government before major launches. They can also give agencies access to advanced forms for up to 90 days before those forms reach users.
Trump lets agencies verify border AI systems before public launch
The AI section of the executive order will emphasize “covered frontier models,” meaning the government will first determine which AI technologies are deemed significant enough to receive additional review. This will not involve an examination at a specific office, but rather at several agencies that will evaluate the forms before issuing them.
The expected executive order will contain two main provisions, one related to cybersecurity and the other related to advanced artificial intelligence models.
While the cybersecurity provision will target the Pentagon, national security agencies, hospitals, financial institutions, and other critical infrastructure across the country, the second provision is about expanding the pool of designated cyber experts.
This includes increasing the number of employees in the US Technical Force – a program launched by PMO Director Scott Kupor late last year. According to Scott Kubor, the purpose of the American Tech Power program was to recruit top-notch artificial intelligence specialists into federal agencies.
The order would also prompt AI companies and the government to share more details about security breaches. This part is about speed. If a company finds a vulnerability or is attacked, federal teams want that information faster, not three meetings and an inbox later.
The Treasury Department is building a clearinghouse while the NSA is getting final review authority for artificial intelligence
The Treasury will lead a voluntary project with AI companies and critical infrastructure owners. This project would establish a clearinghouse within 30 days. The clearinghouse’s mission will be simple: find vulnerabilities and help fix them.
The Office of the National Cybersecurity Director, the National Security Agency, and the Cybersecurity and Infrastructure Security Agency will support Treasury’s work. CISA and the National Institute of Standards and Technology will also help build the model review process.
The second section gives Treasury, CISA, and NIST 60 days to establish a confidential testing process to determine what would be considered a covered border model.
White House Chief of Staff Susie Wiles, National Cyber Director Sean Cairncross, and Michael Kratsios, who leads the White House Office of Science and Technology Policy, will also participate. Thereafter, Susie, Sean and Michael remained connected to the operation through their offices.
The NSA will have the final say after talking with other agencies. The White House began meeting with technology and cyber groups after presenting Anthropic Mythos last month to a small group of technology companies and security researchers.
A White House official called Reports ‘Speculation’ He said any real announcement would come from Trump. The wording also showed disagreements within the Trump administration over how much review frontier AI models should undergo before they are launched.
