Anthropic pitches AI safety bill that may block a model if 50+ people could be killed
AI-summarised brief · reviewed before publication
Anthropic has drafted a federal AI‑safety bill that would allow the U.S. Commerce Department to block the release of an artificial‑intelligence model if an “imminent, foreseeable” scenario could kill more than 50 people or cause over $1 billion in property damage. The bill defines such catastrophes as stemming from a biological weapon, a model evading its developer’s controls, or a cyber‑attack. Independent evaluators, registered with the Commerce Department and paid by AI firms on a non‑contingent basis, would assess risk and, within 72 hours of a dangerous finding, notify the department, which must then seek a judicial order to halt distribution. Federal agencies would be barred from acting unilaterally. The proposal mirrors California’s Senate Bill 53 and reflects Anthropic’s broader “Advanced AI Framework” advocated to lawmakers since June.
💡 Why It Matters
- · By tying model shutdowns to judicial approval, the bill creates a legal bottleneck that could slow or prevent the deployment of high‑risk AI systems, forcing the industry to confront safety assessments before market release.
