Exclusive / Bipartisan AI safety talks accelerate in Senate - Semafor
The Senate’s bipartisan AI safety bill is showing some signs of life.
Key lawmakers said in interviews they may even be able to get to a vote in the Senate Commerce Committee before leaving Washington before the midterm elections, even as the House is set to leave town until November on Wednesday.
“I would love to mark it up this month, if we can get bipartisan agreement,” Senate Commerce Committee Chair Ted Cruz, R-Texas, told Semafor. “We’re making progress. The negotiations are ongoing. We’re not there yet, but I think we’re making meaningful progress.”
Sen. Amy Klobuchar, D-Minn., said she’s “continuing to negotiate” the text of a forthcoming bill with her interested colleagues though she said that President Donald Trump’s social media posts railing against AI regulation aren’t “helpful.”
“This bill is focused on getting approvals, presenting your models to government experts, [and] alerting of imminent harm, like we’ve seen recently. I continue to believe we could get this done. We’d have to put forward the bill, and get support,” Klobuchar said.
The top Democrat on the Commerce panel, Sen. Maria Cantwell, D-Wash., said she was working with Klobuchar on the bill. Both Klobuchar and Cantwell said the proposal is just one part of a larger legislative effort that they want to see done on AI but argued Congress needs to act now on urgent potential threats posed by AI and leave the bigger regulatory swings for later.
“We think it’s part of the equation, we don’t think it’s the whole story. It’s getting the federal government to start playing a larger role and being more prepared and understanding the risk assessment landscape and what the federal government should be doing in that environment,” Cantwell said.
Cantwell and Klobuchar said Cruz could hold a committee vote on the bill soon, according to the Democratic senators. Senate Majority Leader John Thune, who is directly involved in the efforts, was more circumspect, saying it “remains to be seen” whether the proposal would ultimately get a committee vote. Its introduction and possible committee action are still under discussion, according to people familiar with the plan.
“Because it is targeted at these immediate risks, it is something we could proceed on sooner rather than later,” Klobuchar argued.
OpenAI’s Chief Global Affairs officer Chris Lehane had an “encouraging” conversation with Klobuchar Tuesday, an OpenAI spokesperson said.
Lehane told reporters Tuesday that the company was prepared to rally around any federal AI safety effort that could pass and that perfect should not be the enemy of good.
Anthropic previously told Semafor the company has been offering feedback on the bill and hasn’t taken a position on it yet.
Despite the optimistic talk from senators, some people familiar with discussions about the proposal say there’s much more to be done. One source told Semafor the remaining sticking points include internal Senate politics, dynamics with the White House, and garnering support from the AI safety community — all big factors that could individually sink the effort.
