Americans Grimly Eye AI's Existential Threat

When you ask ordinary Americans about the future of artificial intelligence, the answer often sounds more like a scene from a sci‑fi thriller than a tech briefing. A new Politico poll, which interviewed over 2,000 adults just this September, found that about 66 % of respondents believe there’s at least a “moderate” chance that advanced AI could wipe out humanity.

What’s striking is how evenly this fear spreads across the political spectrum. Roughly 70 % of voters who backed Kamala Harris in the 2024 election say AI is a moderate‑or‑higher existential danger, while about 60 % of Trump supporters share that worry. The gap isn’t huge, but it hints at a rare moment of bipartisan anxiety.

Even more unsettling, 17 % of the whole sample said they are “almost certain” AI will end us all — a sentiment that splits almost evenly between Harris and Trump voters. On the other end of the spectrum, only 22 % think AI poses a very small or no risk at all.

These numbers arrive at a time when the AI community itself is sounding the alarm. Last week, an Anthropic researcher quit, accusing the company of “gambling with our lives,” and saying the engineers building these systems “earnestly believe that it could kill us all.” Not long after, Anthropic’s Dario Amodei published a sprawling essay urging the nation’s top AI firms to collectively slow down development — a plea that caught endorsements from Elon Musk and OpenAI’s Sam Altman.

Despite the vocal few, most Americans seem to take industry warnings seriously. Only about a quarter (26 %) of respondents think AI leaders are exaggerating existential risks for marketing flair. In contrast, a solid 48 % say AI is already advanced enough that we should hit the pause button on further progress.

Even here, politics colors opinion. Among Trump voters, 44 % favor a pause, versus 58 % of Harris supporters. Conversely, 40 % of the latter group argue we should keep pushing forward to reap AI’s benefits, compared with 28 % of the former.

So the picture emerging is one of broad consensus that AI could be dangerous, but a divided stance on the right way forward. Whether policymakers will heed the public’s unease or the industry’s own self‑regulation remains to be seen.