American teenagers are using AI heavily, but not the way the doom-and-hype cycle suggests. A Pew Research Center survey published in February 2026 found that 57% of US teens had used chatbots to search for information, 54% had tapped them for schoolwork, and 47% had used them for fun or entertainment. Only 12% had used them for emotional support or advice — a low number that cuts against the narrative that Gen Z is offloading its inner life to ChatGPT.
Pew's headline finding, restated by MIT Technology Review's reporting, is that teens are more than four times more likely to use AI in innocuous ways than potentially harmful ones. That framing matters because most policy debate around teen AI use assumes the opposite. The actual usage pattern looks closer to how earlier generations used CliffsNotes or programmed formulas into a TI-82 — a shortcut layer on top of school, not a replacement for social life.
The interviews MIT Technology Review conducted with kids aged 10 to 18 add texture the survey numbers can't. Some teens are enthusiastic builders. Others refuse to touch the technology on environmental grounds. A significant share told reporters they find the whole category boring, which may be the single most underreported fact in AI coverage this year.
Key facts
- 0157% of US teens have used chatbots to search for information and 54% to help with schoolwork, per Pew's February 2026 survey.
- 02Only 12% of US teens report using chatbots for emotional support or advice, undercutting the dominant AI-companion narrative.
- 03Teens are more than 4x more likely to use AI in innocuous ways than potentially harmful ones, according to Pew.
- 04Princeton voted to let faculty proctor exams for the first time since its 1893 honor code, a change driven by AI cheating concerns.
- 05In-class English essays are now capped at 70 minutes at some schools as teachers respond to AI-assisted writing.
Remy, a 16-year-old in New York, uses the free version of Claude to write programming utilities and has tested AI for game development. His verdict on the state of the art is unsentimental: the code is sloppy, the outputs need correction, and he'd rather build things himself. He compared today's models to the first car or the first airplane — interesting but crude.
Schools are already restructuring around AI use. Remy noted that his English class now does 70-minute in-class essays because teachers assume take-home work will be generated. Princeton voted to let faculty proctor exams for the first time in over a century, ending an honor-code tradition that dates to 1893. Both changes are downstream of the same problem: teachers cannot reliably distinguish student writing from model output, so the assessment surface is moving back into the classroom.
The security stories teens track are also more sophisticated than adult coverage assumes. Remy cited a case in which attackers used Instagram's chatbot to trigger password resets on high-profile accounts, including one belonging to Obama, by convincing the bot to send reset emails to attacker-controlled addresses. That's a prompt-injection-adjacent attack chain, and it's the example a 16-year-old reached for when asked what surprised him about AI.
“Overall, I think AI right now is sort of like the first car or the first airplane. It's interesting but crude. It's obviously an amazing invention but not actually good yet.”— Remy, 16-year-old student, New York
Danielle, an 18-year-old in California studying engineering, is building Next Voters, a multi-agent system that scrapes city council and state government sources, verifies them, categorizes them, and produces a weekly newsletter with cited plain-language summaries. One agent finds official sources, another verifies credibility, another scrapes weekly, another categorizes by topic, and a final agent writes the newsletter. It's a working production pipeline built by a high-school senior.
Her worry is not job displacement. It's epistemic. Danielle told MIT Technology Review she is most concerned about erosion of critical thinking and the spread of misinformation, disinformation, and bias through excessive trust in AI outputs. That framing — AI as an amplifier of existing information problems rather than a jobs threat — is closer to how researchers talk than to how cable news does.
“AI isn't inherently good or bad; that's decided by the people using it.”— Danielle, 18-year-old student, California
The environmental objection is also live among teens in a way it isn't among adult users. Hazel, a 17-year-old rock climber in New York who wants to become an ecologist, said the data centers behind AI use large volumes of water drawn from local supplies and can raise local temperatures. Her one-word description of how AI makes her feel was 'angry,' and her advice to other teens was to stop using it. Whether or not the specific facility she has in mind is closed-loop or air-cooled, the political stance is now part of how a subset of teens defines their relationship with the technology.
Wesley, a 14-year-old in Ohio, drew a line that many adult knowledge workers would recognize: technical help like grammar and spelling checks is fine, but generation of creative work is not. He wrote a short story and ran it through ChatGPT for proofreading, and used it to debug a small game he'd coded. Creativity stays human; mechanical assistance is welcome. That is roughly the split most professional writers and coders are settling into.
Adults writing AI policy for teens tend to assume two failure modes: cheating and emotional dependency. The Pew numbers make the first real and the second much smaller than assumed. The interviews suggest a third pattern nobody is legislating for — a cohort of teens who have already decided AI isn't good enough, isn't ethical enough, or isn't interesting enough to build their lives around, and who are quietly opting out.
For the AI market, the implication is uncomfortable. The consumer-AI thesis assumes today's teens become tomorrow's power users, locked into whichever assistant they grew up with. The MIT Technology Review interviews suggest the actual pipeline is more fragmented: some teens will be sophisticated builders on Claude and ChatGPT, some will be light utility users, and a meaningful minority will be principled abstainers. Retention math built on the first group alone will overstate the addressable market. The companies that win the next decade of consumer AI will be the ones whose products a skeptical 17-year-old is willing to open in the first place.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
