OpenAI recently released a policy document calling on the U.S. federal government to establish mandatory third-party evaluations for advanced AI models. The proposal directly conflicts with President Trump’s executive order on June 2, which prohibits mandatory licensing for AI models and only encourages a voluntary industry-led safety review framework.
On the regulatory structure, OpenAI advocates for oversight by civilian agencies rather than the defense sector. This diverges from the Trump administration’s plan to distribute AI risk assessment responsibilities to the Department of Defense, NSA, and other national security bodies. OpenAI stressed in the document that voluntary cooperation is no longer sufficient to manage the risks posed by advanced AI. Consequently, CEO Sam Altman plans to meet with White House officials and members of Congress in the near future to discuss the regulatory disagreements.