Three out of four Americans say AI firms not doing enough to prevent disaster, Reuters/Ipsos poll finds | Reuters
A Reuters/Ipsos poll finds 39% of Americans think AI has a negative effect on society, the highest share recorded so far. A majority also back slowing AI development, and most prefer safe, responsible AI even if it means the US falls behind globally.
- 39% say AI has a negative impact on society, the highest share the poll has recorded.
- 55% say slowing AI development would be good, while only 13% say it would be bad.
- 73% prioritize safe, responsible AI over keeping the US ahead of other countries.
Three in four Americans think AI companies aren't doing enough to prevent disaster, according to a Reuters/Ipsos poll. The headline numbers are stark: 39 percent say AI has a negative impact on society, the highest share the poll has recorded on that question. Fifty-five percent say slowing AI development would be a good thing, while only 13 percent think it would be bad. And when asked to choose, 73 percent say safe and responsible AI matters more than keeping the United States ahead of other countries in the race. Those three figures together paint a public that is uneasy and leaning toward caution, even at the cost of competitive advantage.
What makes this worth reading is the gap it exposes. Americans are not simply anti-technology; the poll points to a specific worry that the companies building these systems are moving faster than their safeguards. The "disaster" framing in the headline reflects that anxiety about catastrophic outcomes rather than the everyday annoyances of chatbots or job disruption. For anyone following AI policy, that distinction matters, because it suggests public pressure may focus on safety testing, oversight, and accountability rather than on banning products outright. The poll also shows that concern about the national rivalry with China, a frequent argument for speed, does not override the desire for caution among most respondents.
The practical takeaway is that the political mood on AI appears to be souring at the same time industry and government are pushing hard for deployment. A nearly three-to-one preference for slowing development over accelerating it is a significant signal to lawmakers, and the rising negative sentiment year over year suggests this is a trend rather than a blip. Readers tracking regulation, investment, or the company-level debate over safety commitments should treat these numbers as a benchmark for how the public will likely judge the next round of decisions.