House Democrats Push AI Safety Vote as Two-Thirds of Americans Fear Existential Risk

Nearly two-thirds of Americans believe artificial intelligence poses at least a moderate risk of destroying humanity, according to a new POLITICO poll released Wednesday, as House Democrats intensify pressure on Republican leadership to advance AI safety legislation before a seven-week recess.

The survey, which found that almost half of Americans favor pausing development of more advanced AI models, underscores growing public anxiety that has yet to translate into congressional action. More than 100 House Democrats are now urging GOP leaders to keep the chamber in session until lawmakers pass "meaningful, bipartisan AI safeguards," POLITICO first reported.

Why it matters

The polling data reveals a significant disconnect between public concern about AI's existential risks and the legislative response. With the 2026 midterms approaching and constituents raising AI safety concerns at town halls, lawmakers face mounting pressure to demonstrate they can govern emerging technology — even as partisan divisions and technical complexity stall progress on comprehensive regulation.

Key lawmaker positions shift

Rep. Ted Lieu (D-Calif.), the No. 4 House Democrat and co-chair of the House Democratic AI Commission, called Wednesday for the Energy and Commerce Committee to hold a markup of the bipartisan FRONTIER Act — the only comprehensive AI safety bill with cross-party support. It marks the first time Lieu has publicly backed moving the legislation forward, though he stopped short of endorsing its contents.

"I absolutely believe that the Energy and Commerce Committee needs to do a markup of that bill, but we're going to go on vacation in a couple days because that was Speaker [Mike] Johnson's [R-La.] choice," Lieu said at a news conference.

The FRONTIER Act, co-authored by Rep. Lori Trahan (D-Mass.) and Rep. Jay Obernolte (R-Calif.), would establish tiered requirements for AI labs including mandatory incident reporting, independent audits, and risk-management frameworks. Trahan defended the bill as a realistic compromise that could actually become law, noting that current voluntary disclosure practices left the public in the dark about this summer's Hugging Face security incident.

Timeline remains uncertain

House Energy and Commerce Chair Brett Guthrie (R-Ky.) declined Wednesday to commit to a committee vote timeline, suggesting action may not come until 2027. "I wouldn't want to do something in a lame duck session to do it quickly and not get it right," Guthrie said during the POLITICO Decoded Summit, adding that if he retains his chairmanship next session, "that would be more of the target."

Obernolte had previously hoped for a September vote, then pushed for November action. The delay reflects both technical complexity and internal party divisions — Democrats remain split over whether to impose guardrails on AI development or ban certain advancements outright, while also disagreeing on federal preemption of state-level AI regulations.

Partisan divide emerges

The POLITICO survey also revealed growing partisan differences on AI regulation. Fifty percent of 2024 Trump voters said the U.S. should not allow itself to fall behind other nations in AI development, suggesting resistance to regulatory approaches that could slow American competitiveness.

Meanwhile, the Democratic AI Commission continues developing its own policy recommendations with no defined release date. "We're still taking input from stakeholders," Lieu said, deferring to House Democratic Leader Hakeem Jeffries on timing.

Sen. Richard Blumenthal (D-Conn.) noted at a Tuesday event that constituents are discussing AI "in a way they had not done before," reflecting the shift in public awareness captured by the new polling data.

The details were first reported by POLITICO.