A bipartisan House proposal known as the FRONTIER Act has gained the support of OpenAI, which is pushing for mandatory safety evaluations of advanced models. The company specifically endorsed requirements for leading AI developers to embed independent outside evaluators to verify the safety of their systems before release. OpenAI also supports congressional efforts to curb the potential for artificial intelligence to be used in creating biological weapon threats.

CEO Sam Altman argues that no amount of American competitive pressure justifies recklessness and calls for a federal framework to manage frontier risks. OpenAI currently formulates explicit safety cases before reinforcement learning runs that significantly increase capability, a practice the company believes should become an industry standard to ensure alignment and monitoring keep pace with model progress.