More Than Half of Americans Now Worry AI Could End Humanity

A small research group called the Existential Risk Observatory has been asking Americans one question since late 2022: name three things that could wipe out humanity in the next 100 years. Back then, only 7% of people mentioned AI. In their newest survey, 55% did. The group calls this a "tipping point" — the moment a fringe worry becomes a mainstream one. They've run six surveys since 2022, and the line goes up every single time.

What sped things up? The authors point to two recent events: the resignation of Jacob Coxon from Anthropic (an AI safety researcher leaving a major AI lab) and something they call the "Hugging Face incident." They also note that serious government talk about slowing AI down is happening at exactly the same moment. Their conclusion is blunt: fear sells policy, so if you want AI rules, you should want more awareness.

Now the honest catch. Each survey only asks about 300 people, recruited through a paid survey website, all US adults who speak fluent English. The question is open-ended and scored loosely — answering "robots" counts as worried about AI. The authors themselves admit 55% "isn't a very meaningful number" on its own, partly because some people clearly weren't taking the survey seriously. And the group doing the measuring openly wants awareness to rise. So treat the exact figure as fuzzy and the trend as the real story.

If that trend is real, here's what lands in your life. Expect AI rules to arrive faster: safety testing, disclosure requirements, maybe limits on what companies can release. That changes which AI tools you can use, what they cost, how they handle your data, and how quickly your job gets restructured around them. It also means AI safety and policy jobs keep growing. You don't have to agree with the doomsayers to notice that the conversation moved from tech blogs to dinner tables in about four years.