President Donald Trump is facing an unusually public break with his own party in Congress over AI policy, after dismissing the week’s safety warnings as a “HOAX” and “conspiracy” while lawmakers from both parties prepare a stack of guardrail bills.
The split follows days of escalating alarm in Washington. Three Anthropic researchers warned last week that advancing AI could pose a catastrophic risk within a decade, and several AI companies disclosed that their models had hacked outside systems during safety tests. Trump answered on Monday, September 14, insisting on social media that “the only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT”. His post drew immediate criticism from lawmakers of both parties.
The bills on the table
The response from Capitol Hill is unusually varied for an issue that has produced more hearings than laws. Republican Senator John Kennedy of Louisiana says he will propose a bill on Wednesday requiring every AI company to build a kill switch into its models — company-controlled, not government-run. Bernie Sanders has introduced a bill to ban work on “artificial superintelligence” and is holding a private briefing for senators this week. Congressman Jay Obernolte, the lead Republican negotiator on a bipartisan framework, is among those working on alternatives.
Democrat Don Beyer urged Congress to cancel its midterm break to work on the legislation, telling The Guardian that “the industry shouldn’t be allowed to regulate themselves; history doesn’t support that at all”. Obama, in a rare intervention, welcomed the debate but wrote on X that “voluntary standards made by a handful of tech companies won’t be enough” and urged Washington to produce concrete laws addressing safety, jobs and “our kids”.
The polling behind the push
Surveys by the University of Maryland’s Program for Public Consultation, cited by The Guardian, found 85% of Democrats and 79% of Republicans want a new federal agency to monitor AI, and 82% and 78% respectively favour government safety tests for AI making critical decisions. That rare bipartisan agreement has not yet translated into a shared bill: some lawmakers fear a slowdown would hand China the lead in the AI race, and others, like Republican Senator Rick Scott, argue companies building “a product that they think can destroy mankind” should be accountable without new regulation.
The debate lands in the middle of a crowded field. Anthropic CEO Dario Amodei’s call for labs to slow the frontier has drawn both support and suspicion — with advocates warning that industry-designed oversight could undercut stricter government rules — and OpenAI has separately been lobbying UK lawmakers to act fast on regulation while the “political window” is open. Senator Chuck Schumer has asked the administration to give senators a classified briefing on AI and to block China’s access to advanced US chips.
Whether anything passes is another question. Florida Democrat Jared Moskowitz was blunt about the odds: “I think Congress has already shown they don’t have the capacity.” The midterms are weeks away, and data centre protests have already made AI a campaign issue. For now the fight is over who sets the rules — the companies building the models, the president who calls the warnings a hoax, or a Congress that polling suggests voters want to act.
Sources: The Guardian, Ars Technica
Further reading on Singularity.Kiwi:
- Trump Rejects the AI Slowdown: ‘Negative Forces’ Raising Concerns That ‘Won’t Happen’ — the President’s original dismissal, one day before Congress split with him.
- The Pause Consensus — When Every AI Rival Agrees to Slow Down, Ask Who Benefits — why the lab chiefs’ joint call drew skeptics on both sides.
- UK Lords Revive an AI Kill Switch in the Cyber Security Bill — the kill-switch idea is already in play in Westminster; Kennedy’s bill would bring it to Washington.
