A Harvard-led survey reveals a sharp increase in young people’s use of AI chatbots for mental health advice, raising questions about safety, effectiveness, and the need for regulation amid mental health care shortages.
The number of adolescents and young adults turning to artificial intelligence for mental health advice has risen sharply, according to Harvard researchers who say the trend is now large enough to demand closer scrutiny of both access and safety. In a survey led by Hao Yu, an associate professor at Harvard Medical School and the Harvard Pilgrim Health Care Institute, about 19% of respondents aged 12 to 21 said they had used AI chatbots for mental health support, up from 1 in 8 in the previous year. The researchers said the findings suggest AI is increasingly becoming a first stop for some young people who may struggle to reach a clinician.
That shift is unfolding against a familiar backdrop: rising need and limited supply. Yu said the United States faces a shortage of mental health professionals, especially child psychiatrists, while many young people still do not receive treatment even when they need it. He argued that AI can feel appealing because it is instant, available at any hour and less intimidating than a human conversation. The trade-off, however, is that general-purpose chatbots are not built as clinical tools and may offer advice that is misleading, overly reassuring or simply wrong. More than half of the users in the JAMA study did not tell a parent, doctor or other adult that they were using AI in this way, which researchers said raises further concerns.
The survey also suggests that use is becoming more embedded in daily life. The JAMA paper found that 43% of users relied on AI chatbots at least monthly and that 92% described the advice as somewhat or very helpful. Researchers also found higher use among older adolescents and among girls. In a separate study published in May, Harvard Kennedy School researchers reported no significant link between AI use and later depressive symptoms in a sample of US adults, although they said harmful effects in vulnerable individuals could not be ruled out. That followed an earlier study in January which found that daily or more frequent AI use was associated with greater depressive symptoms among adults, particularly among people using it for personal reasons.
Taken together, the research points to a fast-moving question for parents, clinicians and regulators: not whether young people are using AI for mental health support, but what they are getting back. Other studies published this year found that women, people with higher anxiety and depression levels and those with stronger social support were more likely to use AI for mental health help, suggesting the tools may be reaching people who are already under strain. Yu said the next priority is to test whether chatbot advice is accurate, useful and safe, especially as policymakers in some states have already begun considering guardrails for health-related AI use.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
