A new Politico poll found that around two-thirds of Americans believe advanced artificial intelligence poses at least a moderate risk of eventually destroying humanity, with concern spanning party lines. The findings arrive as Washington sees an unusual degree of bipartisan alarm over AI's potential dangers, even as President Trump continues to reject calls for regulation or a slowdown, dismissing such fears as a "hoax" and prioritising America's competitive position against China.

The poll, conducted by Public First between 13 and 15 September, found 70 per cent of 2024 Kamala Harris voters and 60 per cent of Trump voters saw a risk from advanced AI, with most Americans favouring slower development over rapid progress. Roughly half of respondents believe tech leaders' warnings about catastrophic AI risk are sincere, while only 26 per cent view them as a marketing tactic. Politically, the issue has produced unusual alliances, with Bernie Sanders and Steve Bannon jointly warning about unchecked AI, while Democrats consider forming a congressional committee on AI and senators Josh Hawley and Richard Blumenthal back monitoring legislation. House Speaker Mike Johnson has rejected a moratorium but backs "guardrails" such as independent auditing, while First Lady Melania Trump has promoted AI in schools while urging vigilance.