Teens Who Turn to AI Chatbots for Comfort Report Far More Emotional Problems

A survey of 39,761 Ontario students found that those who used generative AI for emotional support were about twice as likely to meet a clinical threshold for emotional problems as students who did not, University of Ottawa researchers report in JAMA Pediatrics.

The paper, published Aug. 31, draws on the Ontario Health and Peer Relations Study, which surveyed students in grades 4 through 12 across four school boards in Ontario, Canada, between May 2025 and March 2026. Of the students surveyed, 8,408, or 21.1%, said they had used AI for emotional support or personal advice.

In unadjusted comparisons, 57.7% of that group scored above a validated clinical threshold on the Ontario Child Health Study Emotional Behavioral Scales, against 29.2% of students who did not report affective use. Students answered both sets of questions at the same time, so the study cannot establish which came first.

The association was smaller but remained after the authors adjusted for loneliness, for mattering (a student's sense of being valued by others) and for demographic characteristics. Students who used AI for schoolwork showed a much weaker association with emotional problems, one that rose slightly with heavier use, according to the paper.

The authors conclude that seeking emotional support from generative AI is "a unique marker of psychological distress" in children and adolescents, independent of loneliness and mattering, and write that clinical frameworks and digital literacy programs should distinguish it from using AI for schoolwork.

The students had a mean age of 12.8 years, and 19,291 of them, or 48.5%, were girls, the paper states. Girls, gender-diverse students and older students reported affective AI use more often than boys and younger students.