Nearly 1 in 5 young Americans turning to AI chatbots for mental health help, study finds
The number seeking help online is roughly the same as the number seeking help from professionals, study cautions
Nearly one in five U.S. adolescents and young adults have turned to artificial intelligence chatbots such as ChatGPT, Gemini, Character.AI or Meta AI for help when they were feeling sad, angry, nervous or stressed, according to a new RAND-led study that researchers said should prompt urgent conversations among parents, clinicians and regulators.
The nationally representative survey, published June 1 in JAMA Pediatrics, found that 19.2% of Americans ages 12 to 21 — an estimated 8.2 million young people — have used generative AI tools for mental health advice. That is up from 13.1% in a comparable RAND survey conducted a year earlier, an increase of more than 40% in roughly 12 months.
The share of young people relying on chatbots for emotional support is now roughly equal to the share who report receiving counseling from a mental health professional, the researchers said.
Most are doing it in secret. Sixty-three percent of young chatbot users said they had not told anyone — a parent, doctor or friend — that they were seeking mental health advice from AI, and nearly 43% said they consulted a chatbot at least once a month, the study found.
“The speed of growth is attention-grabbing, but so is the fact that most young people who use these tools for mental health advice say they are not telling anyone,” said lead author Ryan McBain, a senior policy researcher at RAND.
McBain called the figures “a sad number, because you’d hope that young people would have the sorts of supportive relationships that they would feel comfortable and empowered reaching out to those around them”
What the study measured
Researchers at RAND surveyed 1,009 adolescents and young adults ages 12 to 21 in November 2025 through RAND’s American Life Panel. The survey did not distinguish between chatbots designed specifically for therapy and general-purpose tools used for emotional support, and it carried a completion rate of 58.4%, which the authors said means findings “contain uncertainty and may be subject to nonresponse bias.”
Use was more common among females than males and more common among 18- to 21-year-olds than among 12- to 14-year-olds. Respondents who had recently discussed mental health with a physician were also more likely to report using AI chatbots for the same purpose.
Ninety-two percent of users said the AI advice they received was somewhat or very helpful — but the researchers cautioned that the rating may reflect chatbots’ well-documented tendency to validate and flatter users rather than the actual clinical quality of the guidance.
A separate RAND, Brown University and Harvard analysis published in JAMA Network Open) in November 2025 had pegged the figure at 13.1%, with rates climbing to 22.2% among 18- to 21-year-olds — suggesting the trend has accelerated as AI tools became more deeply embedded in young people’s daily lives.
A ‘pseudo-relationship’ for vulnerable users
The findings land amid mounting concern from physicians and psychologists that commercial chatbots are not safe substitutes for licensed care. The American Medical Association cited the JAMA Pediatrics study in calling on Congress to act on AI mental health tools used by minors.
The American Psychological Association (APA) has issued a health advisory warning that generative AI chatbots and wellness apps should not be relied on to deliver psychotherapy, urging developers to prevent unhealthy dependencies and to create specific safeguards for children, teenagers and other vulnerable users.
Researchers convened by the National Academy of Medicine have warned that AI chatbots may be uniquely capable of causing harm because they “simulate an actual relationship and its associated emotions” rather than simply serving up information.
Adolescents are especially susceptible to that “pseudo-relationship” because their brains are still developing, the panelists said, and chatbots’ tendency toward unconditional affirmation can be “addictive” and crowd out the difficult conversations and critical thinking young people need to develop, according to the National Academy of Medicine.
The same panel documented instances in which chatbots represented themselves as licensed nurses or therapists, shared information about lethal means of suicide, simulated sexual content with minors and coached young people on how to hide mental health symptoms from adults.
A 2025 Stanford University study similarly found that leading therapy-style chatbots exhibited stigmatizing responses toward conditions such as alcohol dependence and schizophrenia, and in some test scenarios enabled rather than challenged dangerous user behavior.
Multiple lawsuits have alleged that chatbot interactions contributed to the suicides of minors, and Illinois last year became one of the first states to bar AI from making independent decisions in therapy.
Calls for guardrails
McBain said the new findings underscore the case for tighter regulation, including requirements that developers track performance, report serious incidents and continuously evaluate how their tools affect users in high-risk situations.
National Academy of Medicine panelists recommended that developers limit how long minors can converse with a single chatbot, reset chatbot memory daily to prevent harmful ideas from compounding, bar chatbots from representing themselves as licensed professionals, route any signal of distress immediately to crisis services, prohibit the sharing of lethal-means content under any circumstances and ban the use of minors’ data for monetization or personalization.
Maybe a real chat, not a chatbot?
For now, the researchers said, the most important intervention may be the simplest: a conversation.
“Many young people appear to be using AI chatbots for mental health advice privately, without the knowledge of parents, clinicians or other adults,” said co-author Jonathan Cantor, also a senior policy researcher at RAND. “That makes it especially important for adults to start conversations about how AI tools are being used and the role they should and should not play.”
If you or someone you know is in crisis, call or text 988 to reach the 988 Suicide & Crisis Lifeline.