Anthropic proposes AI safety bill to US lawmakers

Anthropic, the developer of AI assistant Claude, has shared a draft bill with US lawmakers, seeking to regulate artificial intelligence technology. The proposal, highlighted by Politico, calls for blocking or halting the release or use of AI models that pose a risk of killing or seriously injuring over 50 individuals, or causing more than $1 billion in damages, provided a judge approves.

The company's representative explained that the bill "empowers the federal government to halt or deter dangerous deployments of the most powerful models." It also requires developers of these advanced models to conduct testing, public transparency, independent evaluations, and implement stringent security protocols. However, the United States has largely relied on voluntary industry pledges and executive orders for AI policy during President Donald Trump's tenure, with no federal AI law in place.

The issue gained attention following security breaches caused by autonomous AI agents, leading to Washington suspending Anthropic's most advanced models for three weeks. AI experts have warned that a group of AI agents could potentially control the entire internet within six to twelve months. The proposed legislation would also supersede certain existing state laws, like those in California.

Anthropic's CEO, Dario Amodei, has called for a slowdown in AI development due to safety concerns. While some, like Charlie Bullock of the Institute for Law & AI think tank, found the proposal "pretty reasonable," it was deemed "weaker" than the FRONTIER Act, another bill introduced in the House of Representatives this summer.