Trump Administration Weighs Shift on AI Regulation With Prerelease Model Reviews

The Trump administration had centered its AI policy on easing regulation and accelerating industry innovation. It is now moving toward a prerelease review system amid concerns that frontier models could enable cyberattacks and pose national security risks. The White House has held talks with Anthropic, Google, OpenAI, Microsoft and xAI, among others. The policy shift could affect U.S.-China AI competition and technology companies’ research and development strategies.

The latest executive order asks companies to voluntarily submit frontier AI models to the National Security Agency and the Cybersecurity and Infrastructure Security Agency for testing 30 days before their public release. Google DeepMind, Microsoft and xAI have agreed to provide unreleased models. The administration will also discuss investments in AI companies this week, but it has not disclosed the meeting date, the size of any equity stakes or potential deal terms.