Anthropic Lobbies U.S. Congress to Advance Frontier AI Safety Legislation
nashnova research
Anthropic is quietly pitching a draft AI safety bill to U.S. lawmakers, built around hard numeric thresholds and judicial review to define when and how the government can intervene on AI models — a framework that could shape frontier-AI regulation for the entire industry.
When can the government step in?
The draft sets two hard triggers: a model must pose a foreseeable, imminent risk of more than $1 billion in damages, or 50 or more deaths or serious injuries.
Numbers alone are not enough. The risk must also stem from one of three scenarios: creation or release of bioweapons, a model escaping developer control, or a cyberattack.
This means → the vast majority of AI models would never trigger this mechanism. The bill confines government authority to the extreme-risk end of the spectrum.
Who decides whether a model is dangerous?
The draft creates independent assessment bodies, required to register with the U.S. Department of Commerce and conduct regular evaluations of labs and their frontier models.
A key safeguard: labs may pay assessors, but compensation cannot be tied to the outcome. In plain terms = you can hire someone to audit you, but you cannot pay for the answer.
If an assessor determines a model poses an "imminent listed risk," it must report to Commerce within 72 hours.
How tightly is government power constrained?
Commerce may petition a judge to halt a model, but the draft explicitly prohibits any other U.S. government department, agency, or official from suspending a model's development, deployment, or internal use.
Only a U.S. District Court judge in the District of Columbia can issue a suspension order. This means → the executive branch cannot unilaterally "pull the plug" — it must go through the judiciary.
This reflects the draft's core logic: balancing safety oversight against constraining the expansion of executive power.
Can this framework actually become law?
According to Politico, citing three people familiar with the matter, Anthropic representatives have been quietly briefing lawmakers on the draft to advance the legislative process.
Neither Anthropic nor the Commerce Department responded to requests for comment.
Whether the framework can secure bipartisan support in Congress is the key variable for the bill's future — it remains a draft, and formal legislation is still a long way off.
市场有风险，内容仅供研究参考，不构成投资建议。