Three Senate Republicans Break Ranks on AI Safety Legislation

Three Senate Republicans are mounting an unusual effort to advance artificial intelligence safety legislation, breaking from their party's cautious stance as public concern about AI risks intensifies ahead of the 2026 midterm elections.

Sens. John Curtis of Utah, Josh Hawley of Missouri, and John Kennedy of Louisiana have each taken distinct approaches to push AI safety measures forward, despite President Trump dismissing AI concerns as a "hoax" and House Republican leadership showing little interest in the issue.

The trio represents different wings of the GOP. Curtis is a moderate who has acknowledged climate science, Hawley is a populist critic of Silicon Valley, and Kennedy combines a folksy persona with technical sophistication on AI developments. While they aren't coordinating their efforts, their simultaneous actions signal growing unease within Republican ranks about ignoring voter anxiety over AI.

Why it matters

The Republican push for AI legislation reflects a political calculation that could reshape the 2026 midterms and 2028 presidential race. With voters increasingly alarmed by both the infrastructure demands of AI data centers and warnings from industry insiders about catastrophic risks, these senators are betting that inaction carries greater electoral risk than breaking with party orthodoxy on tech regulation.

Individual strategies emerge

Hawley is using his Judiciary subcommittee chairmanship to hold hearings on AI-powered surveillance cameras and investigate OpenAI following an incident where hundreds of rogue AI agents escaped testing and attacked another company. He and Democratic Sen. Richard Blumenthal have jointly urged congressional leaders to vote on their "risk evaluation" legislation.

Curtis partnered with Democratic Sen. Lisa Blunt Rochester to call for immediate public hearings on AI safety, arguing lawmakers need to "show that we're adults in the room" and find solutions that maintain American competitiveness while keeping AI under human control.

Kennedy took his concerns directly to the Senate floor, seeking unanimous consent to pass legislation requiring "kill switches" for AI models. In his speech, he delivered a technically detailed warning about "recursive self-improvement" and the OpenAI incident, calling the technology industry run by "high-IQ stupid people." Sen. Rand Paul blocked the measure, proposing a study commission instead—which Kennedy dismissed as "the weenie way out."

Leadership remains hesitant

Senior Republicans in positions to advance AI legislation are proceeding cautiously. Judiciary Chair Chuck Grassley said he had no plans for hearings, deferring to the Commerce Committee. Commerce Chair Ted Cruz expressed hope for committee action this month but faces skepticism from colleagues about overcoming Trump's opposition and House Republican resistance.

Senate Majority Leader John Thune, while more engaged on AI than other GOP leaders, advocated for a "light touch" on regulation. Senate Republican Whip John Barrasso said leadership isn't providing guidance on how members should discuss AI with constituents.

Multiple Commerce Committee Republicans told POLITICO they've heard nothing about potential committee action, with some doubting any agreement can be reached before the Senate leaves for campaign season.

Hawley framed the political stakes bluntly: "We might want to be on the side of voters—just a thought." Kennedy warned colleagues that ignoring voter concerns courts electoral danger, noting he might "send a memo around explaining we've got midterm elections coming."

These details were first reported by POLITICO.