US Senate Weighs 'Duty of Care' for AI Developers in New Bill
Senate negotiators are debating a bill that would require AI developers to prevent catastrophic risks, let Washington block unsafe models and bar states from enforcing certain AI laws.
US Senate negotiators are working on legislation that would place a legal responsibility on technology companies to design safe artificial intelligence products, with federal courts playing a role if the government moves to block the release of a model.
The proposal would establish a "duty of care" for AI developers, requiring them to build products aimed at preventing catastrophic risks. It would apply to the most advanced AI models, a category that includes systems from Google, Anthropic and OpenAI.
Under the discussions, Washington would gain the power to stop the release of certain models deemed unsafe, though the exact scope of that authority is still being negotiated. Companies would be able to contest such a decision in federal court.
The measure would also prevent states from enforcing their own laws covering certain AI-related risks. One example cited was the possibility that a bad actor could use an AI system to design nuclear or biological weapons.
The talks involve Senate Majority Leader John Thune, Senate Commerce Committee Chairman Ted Cruz and Senator Amy Klobuchar, the leading Democrat in the negotiations. Senator Maria Cantwell, the top Democrat on the Commerce Committee, has also weighed in.
Cruz said in a post on X that he is working with Klobuchar and Thune on legislation to address catastrophic risks involving biological or nuclear threats. Klobuchar said she is continuing to pursue a bipartisan agreement on government oversight of the greatest risks posed by AI models, including requiring developers to work with government experts to verify and test models.
Thune's spokesperson declined to comment on the negotiations.
The discussions come amid growing pressure on Congress and President Donald Trump to address the risks of advanced AI systems. Concerns have intensified after reports of AI agents deviating from human instructions and hacking external systems, along with public warnings from researchers who left leading companies citing worries about the technology's dangers. Lawmakers from both parties have called for greater oversight in recent days.
Even if negotiators reach an agreement, the congressional calendar could complicate passage. The House of Representatives is scheduled to be in session for only one week before the Nov. 3 midterm elections, while the Senate is expected to be in Washington for three weeks.