Sixty-seven percent of American children between the ages of 9 and 17 have used an artificial intelligence chatbot, according to a census released this week by Common Sense Media — and use begins years before the age of 13, the minimum on most of the popular services.
The figure climbs with age: 58% among 9- to 12-year-olds, 71% at 13 to 15, and 77% at 16 and 17. Taken more broadly, 86% of the children surveyed use some form of AI, and 24% use it every day.
The survey was conducted for Common Sense Media by SSRS, which polled 1,204 children aged 9 to 17 between March 18 and March 26, 2026, through a probability-based panel. The margin of error is 3.8 percentage points, and 50 of the 1,204 interviews were conducted in Spanish.
What they use it for
Among children who use AI, the most common uses are entertainment (89%) and schoolwork (85%), followed by making images, videos, stories, music or apps (75%).
The answers further down the list are the ones drawing attention: 57% have asked AI for information or advice about health or their bodies, 49% for advice on future decisions or goals, 40% have used it to practice conversations or social skills, and 37% have discussed feelings or personal problems with it.
ChatGPT is the tool most of them name, at 40%, followed by Google's AI products at 24%. No other service reaches 5%.
A second survey, a higher number
A separate survey by the JED Foundation, a mental health organization, reached 5,500 students aged 9 to 17 in schools across the country and found that 90% had used an AI chatbot. Nearly two-thirds reported using one both for schoolwork and as a companion — or when they felt stressed, lonely, anxious or overwhelmed.
JED also found that students who reported emotional distress were more likely to report poorer mental health and well-being. Kayla Anderson, a psychologist and the foundation's vice president of research and evaluation, said the data show a correlation rather than a cause: it is not clear whether chatbot use contributes to distress, or whether children already in distress are more likely to turn to a chatbot.
Anne Maheux, a developmental psychologist at the University of North Carolina at Chapel Hill who was not involved in either survey, called the findings "remarkable," pointing to how quickly the behavior has spread. She told NPR that early research suggests the relationship may run in both directions. In a forthcoming UNC study of middle school students in one state, those who used chatbots for companionship grew lonelier over time.
Children still prefer people
The Common Sense data complicate the picture of AI as a replacement for human contact. Asked whom they would turn to first with a feeling or a personal problem, 5% of children named a chatbot, against 56% who named a trusted adult and 32% a friend. On health questions, 73% would ask an adult first and 12% a chatbot — though among daily AI users that rises to 27%.
Where AI does displace people, it does so for a minority with a strong attachment: of the children who have discussed feelings with a chatbot, 25% say it sometimes understands them better than most people do.
What they understand about it
Two findings sit underneath the rest. Only 35% of the children correctly answered that AI cannot reliably tell true from false, and only 32% knew that an AI system is not responsible for what it says.
Meanwhile, 56% said a parent or guardian has talked with them about using AI safely — 59% among children who actually use it, leaving roughly four in ten users who have not had that conversation at home.
Reporting from Common Sense Media, the JED Foundation and NPR.
