Casablanca – Anthropic has proposed a federal law that would allow US authorities to seek a halt to an artificial intelligence model’s release if it poses an imminent threat of killing or seriously injuring more than 50 people, or causing over $1 billion in property damage.
According to Politico, the draft bill, which the AI company has presented to US lawmakers, sets out specific conditions for government intervention in the development and deployment of advanced AI systems. TechCrunch reported on October 20, citing an earlier report by Politico.
The proposal limits intervention to risks linked to the creation or release of a biological weapon, an AI model evading its developer’s controls, or a cyberattack. The threat must meet the bill’s threshold for a foreseeable and imminent catastrophic risk.
Under the draft, independent evaluators would assess AI companies and their most advanced models. The evaluators would register with the US Department of Commerce and conduct periodic reviews of potential safety risks.
AI companies could pay for these assessments, provided they do not tie payments to the results. If evaluators identify an imminent risk covered by the bill, they would have 72 hours after completing their assessment to report it to the Commerce Department.
The department would then need to ask a federal judge in Washington, DC, to halt the model’s release or use. The draft explicitly prevents other federal agencies and government officials from independently suspending a model’s development, deployment or internal use.
Read also: Anthropic Warns AI Could Soon Start Building Its Own Successors
A spokesperson told Politico that the draft translates proposals from the company’s Advanced AI Framework into legislation. Anthropic has promoted the framework to federal and state lawmakers since June.
The proposed thresholds resemble provisions in California’s Senate Bill 53, which lawmakers recently passed into law. That legislation also addresses catastrophic risks associated with advanced AI systems.
The proposal comes as technology companies face growing scrutiny over AI security. Recent incidents involving AI programs that allegedly disobeyed their operators and attempted to hack websites have raised concerns about whether developers have adequate safeguards in place.
Anthropic’s plan also reflects a broader debate over how much authority governments should have over AI development. Technology companies have often favored voluntary safety measures and independent evaluations over mandatory government oversight.
Other industry leaders have taken a different position on regulation. Nvidia CEO Jensen Huang recently argued that the AI industry did not need new laws, according to TechCrunch.
Anthropic’s draft would establish a formal legal process for restricting dangerous AI deployments, but it would also place limits on when and how the federal government could intervene.
