AI Usage Stats 2026: Americans Use It More but Trust It Less

Four in every hundred American adults now say they reach for an AI chatbot almost constantly. Not daily, not several times a day, but almost without pause, the way you might glance at a clock. It’s a small number. But it sits at the leading edge of a shift that Pew Research Center has just put hard figures to: roughly half of U.S. adults now use chatbots, up from about a third just two years ago.

The rising trend of AI usage indicates a significant shift in how technology is integrated into everyday life. And you’d expect a tool burrowing that deep into daily life to have charmed its users. It hasn’t. Not even close.

That’s the strange double exposure running through Pew’s report, published on 17 June and built on a survey of 5,119 adults conducted in February. Use is climbing fast. Warmth isn’t. More Americans expect AI to harm society over the next two decades than expect it to help, by something like four-in-ten against a far thinner slice. They reckon it’s moving too quickly. They suspect it’ll leak their secrets. And the people doing the most reaching, the youngest adults, turn out to be among the most uneasy.

Start with the reaching itself. People aren’t mostly using these systems for the eerie, intimate things the headlines fixate on. They’re using them as a faster card catalogue. About four-in-ten adults turn to a chatbot to look something up, and among employed adults, 38% have folded one into their work. The flashier uses trail well behind: a quarter or fewer make images, chase medical advice, or ask after diet and fitness. News from a bot? Thirteen percent.

Understanding AI Usage in Daily Life

Then there’s the long tail everyone argues about. One in ten say they use chatbots for emotional support or advice; a smaller share, for plain companionship. A minority habit, but a real one, and it sits at the centre of a live and sometimes heated debate about whether software ought to be in the business of keeping people company at all.

ChatGPT still owns the landscape the way an early arrival often does. A little under half of all American adults, 44%, now say they’ve used it, up from 34% a year earlier and more than double the share when Pew first asked back in 2023. Gemini sits a distant second at roughly a quarter of adults, with Copilot and Meta AI behind it; Grok, Claude and Character.ai gather up the ones and the single digits.

Where the picture splits hardest is along the seam of age. Adults under 50 are about twice as likely as their elders to have used ChatGPT, 57% against 28. You can almost picture the household divide, the teenager treating the thing like a study partner while the grandparent who has heard of it leaves it well alone.

The fluent ones are the most worried

Here’s the twist that ought to give the optimists pause. Younger Americans arn’t the cheerleaders their usage might suggest. They are, if anything, the sceptics. Among adults under 30, nearly half expect AI to harm society over the coming 20 years. And the further down the age ladder you go, the gloomier the forecast runs, even as the usage numbers climb the other way. The generation most fluent in the technology is the least sold on where it’s heading. They’ve spent the most time inside the machine, and that time has not reassured them.

What are they afraid of? Mostly the same thing people have feared about every system that sucks up information: that it’ll turn on them. Roughly seven-in-ten Americans expect AI to make their personal data less secure, against just 3% who think it’ll help. Two-thirds say the whole enterprise is barrelling along too quickly, set against 2% who wish it would get a move on.

That unease carries a sharp political twist. Distrust of the government’s ability to regulate AI is now broadly shared, but the parties have swapped seats getting there. In 2024 it was Republicans who fretted more. By 2026 it’s Democrats: 74% of Democratic-leaning adults now say they’ve little or no confidence in the government to regulate AI effectively, a jump of roughly 20 points, while Republican doubt has eased from 70% to 61. The thing being feared hasn’t moved much. Who people trust to handle it has flipped, tracking, perhaps, the change in who holds power.

Meanwhile the technology keeps furnishing the house. A third of adults now own a smart speaker, the Echo on the counter or the HomePod by the bed; roughly one in five has a video doorbell minding the porch; smartwatches ride on nearly four-in-ten wrists, counting steps and pulses. And six-in-ten Americans say they’ve read those AI-written summaries that now crown a page of search results, often without having chosen to. The scepticism in the survey coexists with a steady, frictionless adoption that never quite stops to ask permission.

That’s the real find buried in Pew’s numbers. We tend to assume people take up a technology because it’s won them over, and refuse it because it hasn’t. The American story with AI in 2026 declines to behave that way. People are letting these systems into their searches, their workdays, their living rooms and onto their wrists while telling pollsters, in large majorities, that they don’t trust where it’s all going and wish it would slow down.

Adoption and approval have come unstuck from one another. And if we keep saying yes to a thing we’re increasingly wary of, then the deciding factor was never really whether we liked it. It was whether we could find a way to live without it, and so far, fewer of us seem able to.

Frequently Asked Questions