Senate Ai Safety Bill’s Path Forward Remains Unclear

A bill led by Sen. Amy Klobuchar (D-Minn.) and Senate Majority Leader John Thune on the catastrophic risks of artificial intelligence faces an uncertain future as the parties remain far apart on key provisions, three people with knowledge of the negotiations told POLITICO.

In versions of the legislation circulated over the past week, AI labs would be held a legal duty to mitigate harms their models cause, four people familiar with the text said. The bill has yet to be finalized.

The people who spoke to POLITICO were granted anonymity to discuss private conversations.

Differences over precise language have reemerged just days after former Anthropic employee Jacob Coxon’s warnings that AI “could kill us all by the end of the decade” fast-tracked Hill discussions about potential action on the technology. In a letter earlier Friday, Rep. Sam Liccardo (D-Calf.) urged Speaker Mike Johnson to keep the House in session until Congress passes AI safeguards.

Drafts of the bill set categories of risks that AI companies would have to consider, such as the potential to create biological weapons or launch offensive cyber operations, according to two of the people.

AI labs, upon discovering a model has advanced capabilities, would be required to disclose their findings to the Commerce Department, along with information on what they have done to counter the risks and a way for the secretary to verify that the dangers have been addressed, according to two of the people. The disclosure would have to occur before the model is released to provide the secretary time to review the models.

Under the framework, if the secretary of Commerce believes that a certain risk has not been properly mitigated, the department could pursue an injunction in federal court, according to two people.

Without an injunction, the AI labs would be able to determine whether to release a model and not be subject to a pre-approval regime.

The bill would also preempt some state AI laws, the four people told POLITICO.

Senate Commerce Chair Ted Cruz (R-Texas) has been involved in negotiations, and has publicly called for legislation to ease potential catastrophic risk.

“It is possible to move bipartisan legislation,” he said on ABC’s The View, “but it ain’t easy.”

Sen. Maria Cantwell (D-Wash.), the top Democrat on the Commerce Committee where the bill would be considered, appeared to throw cold water on the negotiations. While she broadly supports regulating AI, she wrote on social media that “the answer is not a weak federal standard that becomes a backdoor for wiping out stronger state protections.”

She also said she would like to see AI models “undergo testing by scientists and experts at our national laboratories.”

Cruz and other Republicans have largely favored a lighter touch approach on AI, warning that overly burdensome regulations could slow U.S. innovation and allow China to dominate the AI industry.

“AI presents great opportunities but also comes with significant risks,” Klobuchar said in a statement to POLITICO, adding that she is continuing to work toward a bipartisan agreement. “That includes requiring developers to work with government experts to verify and test models to make sure AI is safe. We cannot allow the release of dangerous models, including those that evade the control of their developers.”

Thune and Cruz did not immediately respond to a request for comment.