Anthropic's Proposed AI Safety Bill Would Require Court Approval to Block Model Releases

Anthropic has proposed a federal AI safety bill that would make it significantly harder for U.S. government agencies to stop the release of artificial intelligence models, requiring approval from a federal judge before any intervention. The confidential draft, obtained by POLITICO and now circulating among lawmakers, establishes a framework in which the Commerce Department could seek to halt a frontier AI model only under extreme circumstances—when it poses a foreseeable, imminent risk of either killing or seriously injuring more than 50 people, or causing over $1 billion in damages. Critically, even in those scenarios, government action would only be permitted with a court order from a federal judge in the U.S. District Court for the District of Columbia.