Eight million American teenagers are now turning to artificial intelligence chatbots for mental health support, a dramatic increase that has left researchers warning about the hidden dangers of unsupervised use. A new survey from the non-profit research institute RAND, published in JAMA Pediatrics, found that 19.2% of young people aged 12 to 21 — an estimated 8.2 million individuals — have sought mental health advice from AI chatbots. That represents a more than 40% rise in just one year, up from 13.1% in a similar survey conducted a year earlier.
Among those who use the tools, the habit is often entrenched. Nearly 43% reported seeking advice at least monthly, and 5.8% said they did so daily or near-daily. Use was more common among females than males, and among older adolescents and young adults aged 18 to 21 compared with younger teens aged 12 to 14. Black youth were more likely than other groups to engage with the chatbots on a monthly basis. The researchers also found that those who had previously spoken with a doctor about their mental health were more likely to use the chatbots, suggesting that traditional care may not be meeting demand — or that the online tools are being used as a supplement rather than a replacement.
Why teens are turning to AI
The rapid adoption comes against a backdrop of a well-documented mental health crisis among American adolescents. Suicide is the second leading cause of death for this age group, according to Johns Hopkins Medicine, and a significant treatment gap persists. In 2024, 40% of teenagers who experienced a major depressive episode did not receive any mental health treatment, according to data cited in the research briefing. AI chatbots offer a low-cost, always-available alternative that feels private and non-judgmental — qualities that appeal strongly to young people who may be reluctant to speak to a therapist or parent.
Yet the very features that make chatbots attractive may also be fuelling a worrying lack of transparency. The RAND survey found that nearly two-thirds of users — 63% — had not told anyone, including parents, clinicians or other trusted adults, that they were using AI for mental health advice. “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 Ryan McBain, a senior policy researcher at RAND and lead author of the study.
While 92% of those surveyed said they found the chatbots’ advice very helpful, researchers caution that this perception may be misleading. AI systems are designed to be agreeable. “By default, AI advice does not tell people that they’re wrong nor give them ‘tough love,’” said Myra Cheng, a computer science PhD candidate at Stanford University, in a statement earlier this year. “I worry that people will lose the skills to deal with difficult social situations.”
The hidden risks to developing minds
For teenagers, whose brains are still developing, the ability to navigate complex social interactions is crucial. Overreliance on AI for emotional support may hinder the development of real-world empathy, conflict resolution and the capacity to handle disagreement — skills that are forged in messy, human relationships, not in one-sided conversations with a compliant machine. The research briefing notes that AI chatbots can simulate empathy and concern, potentially fostering emotional dependence and a false sense of intimacy that is especially risky for vulnerable individuals.
Beyond the social-skills concern, the risks are substantial and well-documented. AI chatbots systematically violate core ethical standards that govern human mental health practice, according to researchers at Brown University. They lack scientific validation and oversight, and often include inadequate safety protocols. Several investigations have linked AI chatbots to the deaths of American teenagers who died by suicide, according to reports. The tools have been found to be inconsistent in answering questions about self-harm and suicide — sometimes offering dangerous advice or failing to respond appropriately to suicidal ideation.
AI chatbots can also generate misinformation with a tone of confidence, making it difficult for an emotionally vulnerable user to detect errors. There are reports of chatbots reinforcing negative beliefs, delusional thinking, or even encouraging dangerous behaviour. An informal term — “AI psychosis” — has emerged to describe cases where vulnerable individuals experience worsening mental health symptoms due to the chatbot reinforcing unhealthy thoughts. Algorithmic bias has also been identified, with some AI therapy tools showing prejudice against certain mental illnesses, potentially leading users to feel judged or shamed. Privacy is another major concern, as chatbots often collect and store sensitive personal data with limited transparency about how it might be shared.
What parents and professionals can do
Given that most young people are using these tools in secret, researchers and mental health organisations are urging parents and caregivers to initiate open conversations about AI use. The American Psychological Association advises parents to talk to their teenagers about what data chatbots may store, ask about how they use the platforms, set rules together, and take note of whether chatbots are replacing real-life interactions. “Know when to get help,” the APA instructs. “AI can’t handle crises. Get professional support when needed.”
AI literacy is also being called for in schools, with educators encouraged to teach critical evaluation of AI-generated content and the ethical implications of using such tools. Developers are being urged to prioritise adolescent safety through age-appropriate defaults, transparency and human oversight. There is a growing push for regulatory frameworks that would hold AI systems to enforceable standards, similar to those applied to medical devices. Mental health professionals are advised to proactively ask patients about their AI chatbot use, help establish boundaries, and emphasise that these tools are aids, not replacements for human interaction.
For those in crisis, immediate human support is available. In the United States, the 988 Suicide and Crisis Lifeline can be reached by call or text to 988, or online at 988lifeline.org. The service is free, confidential and available 24 hours a day. In the United Kingdom, the Samaritans can be contacted on 116 123 or by email at [email protected].
