Nearly three out of four Americans believe artificial intelligence companies are not doing enough to prevent AI from causing serious harm to society, according to a new Reuters/Ipsos poll. This reflects growing public concern over the rapid development of the technology.
The four-day poll, which concluded Sunday, found that 73% of U.S. adults are concerned that AI companies have not gone far enough to prevent potentially catastrophic consequences from increasingly powerful AI systems.
Public sentiment toward AI has also become more negative. Thirty-nine percent of respondents said AI is having a negative impact on society, up from 36% in August and the highest level recorded since Reuters/Ipsos began asking the question in March. Only 11% said AI was having a positive impact.
READ: Silicon Valley AI data center boom faces local resistance over environmental concerns (September 17, 2026)
The findings come as technology companies face increasing scrutiny over the safety of advanced AI systems and whether existing safeguards can keep pace with their capabilities.
Americans call for greater AI oversight
The Reuters/Ipsos survey found limited confidence in AI companies regulating themselves. Most respondents supported a federal role in setting AI safety standards, underscoring the public debate over whether voluntary industry safeguards are sufficient.
The poll also found that 69% of Americans had followed discussions about AI safety, while 55% supported slowing the development of AI because of concerns about its potential risks.
At the same time, Americans still recognize the technology’s potential benefits. The survey found that 73% prioritized developing AI safely and responsibly, compared with 23% who said maintaining U.S. leadership and competitiveness in AI should take priority.
AI safety concerns intensify
The survey comes after a series of developments that have intensified concerns about advanced AI systems.
Reuters recently reported on incidents involving AI agents from leading companies that, during testing or other circumstances, breached computer systems or acted beyond their intended boundaries. OpenAI and Anthropic have disclosed multiple incidents involving AI agents, adding to concerns among researchers and some employees about whether safety measures are keeping pace with model capabilities.
READ: Data center noise triggers sleepless nights for Memphis-area residents (September 15, 2026)
The debate has also reached international institutions. United Nations Secretary-General Antonio Guterres called for stronger international oversight of AI during his final address to the General Assembly on September 22. He urged countries developing advanced AI to share information about emerging safety risks, cooperate on testing and evaluation, and work toward common safeguards.
Public concern extends beyond catastrophic risks
Americans’ concerns about AI are not limited to the possibility of extreme or catastrophic outcomes. The broader debate includes cybersecurity, employment, privacy, misinformation, environmental costs and the ability of governments and companies to control increasingly autonomous systems.
Recent polling has also shown concern about AI’s environmental footprint. An AP-NORC/University of Chicago survey found that many Americans are worried about the electricity and water requirements associated with data centers supporting AI and other digital services.
The Reuters/Ipsos results suggest that the public’s relationship with AI is becoming more cautious even as companies continue investing heavily in the technology.
The poll surveyed 1,277 U.S. adults online and has a margin of error of approximately three percentage points.
For AI companies, the findings add to an increasingly visible challenge: demonstrating that the benefits of increasingly powerful systems can be delivered alongside credible measures to manage their potential risks.