Key facts
- Bill allows US to block AI models posing imminent risk of 50+ deaths or $1B in damages.
- Risk must stem from biological weapon creation, model evasion, or cyberattack.
- Independent evaluators would assess models and report risks to Commerce Department.
- Commerce Department can seek judicial order to suspend model release.
- Bill preempts some state AI laws on risk disclosure and incident reporting.
Anthropic, a leading AI company, has proposed a draft bill to US lawmakers that would establish a framework for regulating the safety and security of advanced AI models. The confidential draft, obtained by POLITICO, suggests that the government could block the release or use of an AI model if it poses a foreseeable, imminent risk of causing more than 50 deaths or over $1 billion in damages. These risks must be linked to the creation of a biological weapon, the model evading its developer's controls, or a cyberattack. The bill also outlines a process for independent evaluators to assess AI models and report risks to the Commerce Department, which could then seek a judicial order to halt a model's deployment. This initiative comes as Congress grapples with how to regulate the rapidly evolving AI industry, with various legislative efforts underway in both the House and Senate.