A new Reuters/Ipsos poll finds broad bipartisan concern that federal leaders are not responding seriously enough to the risks of artificial intelligence, with jobs, data-center construction and military targeting among voters’ leading worries.
The nationwide online survey, published October 7, collected responses from 4,506 U.S. adults, including 3,526 registered voters. Reuters reported a margin of error of about 2 percentage points.
Because this is an opinion poll, its results measure public attitudes at the time of the survey. They do not establish whether a particular AI system is safe or whether a forecast about jobs will occur.
Majorities want stricter AI rules
Among registered voters, 57% said the Trump administration was not taking AI risks seriously enough or was not taking them seriously at all. A similar 54% said Congress was falling short.
Support for stricter government regulation crossed party lines: 80% of Democrats and 61% of Republicans favored tighter rules, according to Reuters. The survey ended Monday, placing the findings in the final month before the November 3 congressional elections.
The results arrive as lawmakers and technology companies debate whether voluntary commitments are adequate. President Donald Trump said last week that six leading developers agreed to voluntary safety principles and independent reviews, but Reuters noted that the agreement did not identify penalties for noncompliance.
Jobs dominate public concern
The poll found that 84% of registered voters viewed AI as a threat to American workers. Reuters compared that with 74% who viewed competition from foreign workers as a threat and 60% who said the same about illegal immigration.
Those comparisons reflect perceptions, not economic measurements. Researchers and companies continue to disagree about how quickly AI will displace jobs, create new roles or change the tasks performed within existing occupations.
The findings nevertheless show that employment anxiety is now central to the political debate over AI. It is no longer limited to software companies or technology regulators.
Data centers and military use also draw concern
Fifty-six percent of registered voters supported limits on new data-center construction. Support included 69% of Democrats, 55% of independents and 43% of Republicans.
Data centers have become local political issues because they can require substantial electricity and water infrastructure. Their economic benefits and environmental costs vary by project and region, so the poll does not resolve individual permitting disputes.
The survey also found that 57% of voters opposed using AI to determine the target of a U.S. military strike. Reuters said that share had increased from 49% in an August 2025 poll.
That question concerns decision-making authority, not every military use of AI. Systems can also be used for logistics, maintenance, surveillance analysis or defensive cybersecurity, which may raise different policy considerations.
What the poll does—and does not—show
The strongest conclusion is that skepticism about AI governance reaches across party lines. It does not show that voters agree on a specific bill, enforcement agency or technical safety standard.
It also captures attitudes during an unusually active period for artificial intelligence. Recent company announcements have emphasized increasingly capable models and expanded cybersecurity access, including Anthropic’s expanded verification program. Those developments can reinforce both optimism about AI’s benefits and concern about misuse.
Congress has not yet enacted a comprehensive federal AI law. The new poll suggests voters may be more receptive to national rules than the current legislative record indicates, but election results and actual policy choices will determine whether that sentiment produces binding requirements.
Image credit: Igor Omilaev via Unsplash. Image is illustrative.