Daily AI Use Is Rising, and So Is Worry: What the Gallup Numbers Show

68% of Americans who use AI every day say they are worried about it. That result comes from a Microsoft-commissioned Gallup survey reported Sept. 23.

Overall, 74% of Americans reported feelings of worry about AI, and only 36% expect AI to mostly help the country, according to Gallup TechCrunch. The study covers 37 countries so far, with about 1,000 respondents in each country polled between April and July, and Gallup plans to reach 140 countries over time. India, Australia and Malaysia were not part of the results reported so far.

Western countries showed the highest levels of concern, with the United States highest for anxiety. Canada followed the same pattern. There, 29% of those surveyed said they use AI daily, and more than half of those daily users expressed concerns about its impact. Among Canadians aware of AI, 64% said it worried them.

The numbers were different in parts of Asia. Singapore had the highest share of daily AI users in the study at 46%. In Singapore, more than 80% of those aware of AI expect it to improve their everyday lives, and 77% think it will help the country. In China, 35% reported using AI daily, and most expect it to help the country. Gallup also found that people who use AI daily were generally more trusting of its results, especially in high-adoption countries. Use builds some trust in answers, but it does not remove worry.

Other recent U.S. surveys found the same mix of routine use and caution. Rutgers found 27% of Americans use AI every day, while fewer than one in 10 would allow AI to make the final call without human involvement Rutgers. A Reuters/Ipsos poll published Sept. 22 found 73% worry that AI companies have not gone far enough to prevent AI from causing serious harm to society Reuters. As of June 2026, 38% of Americans said the AI coverage they see is mostly favorable, down from 44% a year earlier, according to Ipsos.

Earlier surveys showed similar concern. In June 2025, half of U.S. adults said increased use of AI in daily life made them feel more concerned than excited Pew Research Center. Pew also reported worry about harm to human creativity and relationships. A September 2026 Pew report across 37 countries found continued pessimism about AI and jobs and economic inequality, with Democrats more likely than Republicans to predict AI will lead to fewer jobs over the next 20 years, 75% versus 68%.

Related findings add detail for builders and operators. About 72% of U.S. adults say it is extremely or very important that a doctor tells them if AI is used in their health care. Views of data centers, the large facilities that run AI systems, have turned more negative since January 2026, though many people remain unsure or unfamiliar with them. In a random sample of 10,000 webpages collected in July 2026, one in 10 showed signs of being written or substantially edited by AI.

The broader context here is that adoption is outrunning assent. People will ask a model for drafting, coding help, summarization and search each morning, then tell a pollster they are worried about where this goes. We have seen this pattern before, when the PC, the browser and the smartphone each went through a phase where daily reliance coexisted with complaints about overload, etiquette and control.

In my view, the signal for technology teams is about control boundaries, not slowing deployment. U.S. users are describing a human-in-the-loop default, meaning a person makes the final decision. They want disclosure in high-stakes uses like health care, veto power over automated decisions, and evidence that providers tested for failure modes, meaning ways the system can go wrong, before scaling. Trust in answers can rise with use while trust in institutions lags. Reliability is necessary but not sufficient. Auditability and provenance, meaning the ability to check how an answer was produced and where information came from, and clear recourse matter as much, like keeping a receipt for a decision.

Worth flagging for product work is that mental models differ by place. Where daily use is highest, expectations for everyday benefit are also highest. Where debate centers on job loss and safety rules, worry persists even among heavy users. I have watched my own children move from casual tries to matter-of-fact dependence on these tools for school and work, still pausing to question an answer that sounds confident but looks thin. That habit, skepticism paired with use, may be the durable norm.

In my view, the optimistic read is still available. Worry at this scale is a demand for better systems, not a rejection. Users are asking for AI that explains its sources, declares itself in clinical and civic settings, leaves a human accountable for the final call, and runs on infrastructure communities understand. Build that, and daily use stops feeling like a compromise.