Anthropic proposes bill to block AI models that could cause $1 billion damage or 50 deaths
Anthropic is pitching a confidential draft AI safety bill to U.S. lawmakers, according to Politico. The proposal would let the government block the release of a new AI model only if the model could cause at least $1 billion in damage or kill or seriously injure 50 or more people. The bill limits the trigger to risks arising from the creation or release of a biological weapon, the model evading…
Key points
- Bill would let government block AI models only if they could cause $1 billion damage or 50+ deaths/injuries.
- Blocking applies when risk stems from biological weapons, control evasion, or cyber‑attack, and requires a judge’s order.
- Independent evaluators registered with Commerce must report findings within 72 hours; labs may pay them if independence stays intact.
Independent evaluators, registered with the Commerce Department, would periodically assess Anthropic’s frontier models and could be paid by the lab provided their independence is not compromised. If an evaluator finds the alarming‑sounding risks, they must report to Commerce within 72 hours, after which a federal judge in the District of Columbia would decide whether to halt the model’s release. The draft mirrors California’s SB 5 AI law and states that no other U.S. agency may suspend the model’s development, deployment, or internal use. The move follows broader calls for AI safety, including proposals from Bernie Sanders and a Trump‑backed “Joint Commitment on Frontier Responsibilities.”
The story so far
2 episodes →- Anthropic proposes bill to block AI models that could cause $1 billion damage or 50 deathsthis story
Anthropic wants AI models blocked only if they threaten over $1 billion in damage or more than 50 deaths or serious injuries
techspot.com · 19 October 2026
Loading the full article…
This text was published by techspot.com and written by Rob Thubron. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- Meta's Muse shared a seller's home address after user granted always‑allow permission · 2 src
- Trump administration bans DeepSeek from civilian government systems · 9 src
- US prosecutors allege ex-OpenAI employee had Saudi royal family ties in espionage case · 5 src
- Experts warn consumers about trusting personal AI agents with money and data · 1 src
- Nvidia and Germany sign joint declaration, pledge nearly $10 billion for industrial AI · 2 src
Comments
via GitHub Discussions