Anthropic Proposes AI Safety Legislation Limiting Government Intervention and State Rules

Anthropic is reportedly presenting US lawmakers with draft AI safety legislation that would establish specific conditions under which federal authorities could prevent the release of potentially dangerous AI models, according to Politico.
The proposal would limit government intervention to defined catastrophic-risk scenarios while potentially preempting certain state-level AI regulations.
Supporters could argue that a consistent federal framework would reduce regulatory fragmentation and provide clearer standards for advanced AI developers.
Critics may question whether the proposed thresholds are sufficiently protective and whether restricting state authority could weaken oversight. The draft has not been formally introduced in Congress, and its provisions are not currently binding law.