Anthropic’s proposed AI safety bill only kicks in if 50 deaths or $1 billion in damages are on the line — proposal would effectively make the Commerce Department the only agency capable of stopping an AI model
The draft would set a high threshold for federal intervention in model release or use.
TL;DR
- Anthropic representatives have presented lawmakers with a draft bill governing frontier AI safety and security.
- The reported draft would allow intervention only for a foreseeable, imminent risk exceeding 50 deaths or serious injuries, or $1 billion in damages.
- The text remains a proposal, and lawmakers have not enacted it.
Anthropic representatives quietly presented Capitol Hill lawmakers with a draft bill on frontier AI safety and security, according to reporting first attributed to Politico. Tom’s Hardware and Reuters’ account of that report describe the same proposed intervention thresholds. [1] [2]
Under the reported draft, federal officials could seek to block a model’s release or use only when they judged there was a foreseeable and imminent risk of more than 50 people being killed or seriously injured, or more than $1 billion in damage. These are thresholds described in reporting on a draft, not current statutory rules. [1] [2]
The proposal would shape when government could intervene in development or deployment of advanced systems. The available reports describe Anthropic’s proposal and the draft’s limits; they do not establish that Congress has accepted the approach or that it is the only policy proposal under consideration. [1] [2]
The draft’s full legislative status and the response from lawmakers remain open. Until a bill is formally introduced and considered, the reported thresholds should be treated as Anthropic’s proposed framework rather than binding policy. [2]
Why it matters
This is a direct attempt by a frontier AI developer to influence federal AI governance. The proposal’s high intervention thresholds make the boundary between company-defined safety processes and government authority the central policy question.
Editor's note
Reported draft; not enacted law. Thresholds are attributed to reporting about the proposal.