Biden Urges Schumer’s AI Working Group to Codify Provisions

President Joe Biden told senators Tuesday that his executive order on AI goes only so far, and Congress should work with the White House to incorporate some of its provisions into legislation, Sen. Todd Young, R-Ind., told us Thursday.

Young is one of four members of a bipartisan Senate working group who met with Biden to discuss AI the day after his EO signing (see 2310300056). Senate Majority Leader Chuck Schumer, D-N.Y.; Martin Heinrich, D-N.M.; and Mike Rounds, R-S.D., also attended.

“The president emphasized there are only certain things that can be accomplished in the executive order,” said Young. “And the extent to which we are in agreement on particular facets of that executive order, we want to work together to memorialize those pieces in statute.” Both the legislative and executive branches are trying to “get clarity” on additional legislative possibilities not included in the EO, said Young. The White House didn’t comment.

“We’re slowly figuring it out,” said Heinrich. Biden’s EO is a “step toward legislating in the same space, and there are going to be places where we have quite a bit of consensus and places where we don’t. And we’ll just have to move forward on the places that we do and that we can get done in this Congress.”

Biden during his signing ceremony raised concerns about AI’s role in youth social media addiction and cyberthreats. He called on Congress to pass legislation banning targeted advertising for children on social media. The EO carves out recommendations for how the FCC can coordinate with NTIA on spectrum management, how the FTC can issue rules to address competition concerns and how the Commerce Department can set federal guidelines for agencies and industry.

Rounds told us meeting participants all agreed on the need to create policies that foster, not hinder, AI innovation in the U.S. This is “critical” for the U.S. economy and national defense, he said: “We’ll have some discussions about what’s the appropriate level of regulatory activity, but it should be focused more on making it a better place for AI to develop rather than trying to impose or limit the development. The other piece we [agreed on] is the American people have got to see the benefits, in terms of quality of life improvements that AI can bring.”

The 2024 presidential election increases the urgency to regulate, Senate Intelligence Committee Chairman Mark Warner, D-Va., told us Thursday. “Inherently you can only go so far with” executive action, he said. “You need legislation because you can’t really have enforcement without” it.

Election security has been one of Warner’s priorities on AI (see 2309070059). Thursday, he introduced a bill with Sen. Jerry Moran, R-Kan., that would require federal agencies to incorporate the National Institute of Standards and Technology’s AI risk management framework into their AI-related operations.

“We thought we would add a little common sense to the conversation” and create a standard that’s been widely well-regarded, Moran told us. Warner said to expect a series of bipartisan bills from him related to AI.

There’s a global race to regulate AI, and it might be wise for the U.S. to allow the EU to go ahead first, said Sen. John Kennedy, R-La. “We’re a long way from being able to legislate,” he said. “There may be some benefit to allowing others to first, like the EU, and watch them screw it up.” Kennedy said he’s “not entirely convinced [AI] ought to be regulated. I think the answer to that is probably but not certainly. And it depends on the type of regulation.” Congress first needs to agree on the “dangers” of AI and then talk about enforcement, he said.