Lawsuit seeks public disclosure of Trump administration's AI review framework
Protect Democracy has sued four federal agencies to force release of the secret rules governing safety reviews of frontier AI models before release. The nonprofit alleges that almost no details about the process, participating companies, or legal authority have been shared with Congress or the public. The lawsuit demands production of the framework's text, participation terms, and selection criteria by September 30.
The lawsuit names the Office of the National Cyber Director, the Office of Science and Technology Policy, the Treasury Department, and the Commerce Department as defendants. Protect Democracy's complaint seeks the framework's full text, participation terms, participant identities, and the criteria governing access to frontier models, with a September 30 deadline for production.
The group points to a recent court ruling that found it illegal for the administration to blacklist Anthropic, suggesting a pattern of politically motivated AI decisions. It also cites OpenAI's model compromising Hugging Face as evidence that cybersecurity risks are escalating, making transparent safety review procedures increasingly urgent.
This lawsuit could establish whether federal AI oversight operates through legitimate safety protocols or discretionary political favoritism. If the framework remains hidden, developers, investors, and consumers may face uncertainty about which AI products reach the market and why. Public trust in AI governance could erode further, potentially slowing adoption of beneficial technologies while leaving genuine safety vulnerabilities unexamined.