AI risk warnings belie lack of consensus in Congress - Semafor

The sudden appearance of alignment among AI CEOs about industry regulation over the weekend masks an inconvenient reality for Congress: There is no consensus.

In the House, Speaker Mike Johnson is downplaying his own role in favor of a “partnership” with AI companies, as he told CNN on Sunday.

Across the Capitol, OpenAI remains most supportive of bipartisan Senate Commerce Committee talks that could advance this week, despite skepticism from Anthropic, Democrats, and AI safety groups, who say those negotiations risk letting the industry self-police.

But Saturday’s three-CEO agreement that frontier models are accelerating too fast centered on Anthropic’s plan to do “things every AI lab needs to do anyway,” as our colleague Reed Albergotti wrote — which amounts to self-regulation, with labs likely choosing their own independent evaluators.

“The time to act is now, before it’s too late,” Rep. Lori Trahan, D-Mass., told Semafor.

She urged Johnson to keep the House in session past this week to work on AI, but her own leader is already looking to 2027.

House Minority Leader Hakeem Jeffries told ABC on Sunday that his caucus would gather this week to discuss options.

At a separate closed-door fireside chat with Jeffries last week, former President Barack Obama urged the New Yorker to “put together a framework for a very public conversation” and said he’d prioritize the issue if he were running for president, according to excerpts of the remarks confirmed by Obama’s office.