Trump may be forced to reveal secret rules feds use for AI safety testing

A nonprofit organization has sued four federal agencies over their refusal to disclose the secret framework used to review artificial intelligence models before they are released to the public. The Trump administration established this voluntary review process to assess national security risks in frontier AI models, but has not released details about how it works, which companies participate, or what legal authority officials have to conduct the reviews. The lawsuit argues this secrecy is problematic because decisions about which AI models are approved could be influenced by political favoritism, with no oversight from Congress or the public, and because safety vulnerabilities discovered later could indicate failures in the review process itself. The nonprofit is requesting that all unclassified information about the framework, including its procedures, terms of participation, and selection criteria, be made public by a court order.