A RAND Corporation study released in June 2026 found that 19.2% of U.S. adolescents and young adults between the ages of 12 and 21 have used AI chatbots such as ChatGPT, Gemini, Character.AI or Meta AI for mental health advice, a figure that marks a more than 40% increase from the prior year. The finding underscores how quickly generative AI has moved from a novelty into a default coping tool for a generation navigating a well-documented youth mental health crisis.
A Crisis Meeting a New Technology
The rise in chatbot use for mental health support has unfolded against the backdrop of a years-long shortage of licensed therapists, especially those who work with children and teens, combined with long waitlists and high out-of-pocket costs for traditional therapy. For many young people, an always-available, free or low-cost chatbot fills a gap that the traditional mental health system has struggled to close, offering an outlet at 2 a.m. when no human counselor is reachable.
What the Research Shows
The RAND findings align with parallel research from the American Psychological Association, which reported that more than three-quarters of psychologists say their patients are now discussing artificial intelligence during therapy sessions, whether as a supplement to treatment or, in some cases, a substitute for it. A CNN report on the same trend quoted physicians describing a spectrum of use, from teens using chatbots for low-stakes venting to more concerning cases where young people rely on AI as their primary source of emotional support during genuine crises.
Where the Danger Lies
Axios reported in May 2026 that leading AI chatbots generally avoid giving overtly dangerous responses to direct questions about self-harm, but they still struggle when mental health risk signals emerge subtly or build gradually across a long conversation, exactly the kind of nuanced situation a trained therapist is taught to recognize. The stakes of that gap became starkly clear in January 2026, when Google and Character.AI settled multiple lawsuits, including one involving a 14-year-old who died by suicide after extensive conversations with the platform’s chatbots, a case that has become a reference point in the broader debate over AI safety guardrails for minors.
The Regulatory Response
The backlash has moved quickly into statehouses. Several states have advanced or passed legislation explicitly restricting AI chatbots from acting in a therapist role, reflecting bipartisan unease about unregulated AI filling clinical mental health functions without oversight, licensing or liability comparable to human providers. Despite this wide and rapidly growing usage, oversight remains inconsistent nationally, with no uniform federal standard governing how these tools should handle mental health conversations or protect user privacy.
The Industry’s Response
The AI therapy landscape has begun sorting into distinct camps, according to reporting on the sector: consumer chatbot startups like Slingshot AI building purpose-made mental health products, clinician-facing platforms like Limbic and Lyra Health that keep licensed professionals in the loop, general-purpose AI labs such as OpenAI, and university research groups studying the phenomenon rather than building products. Investment and momentum appear to be shifting toward the clinician-supporting model, in part because it sidesteps the liability and regulatory exposure of unsupervised AI providing therapy-like advice directly to vulnerable users.
What’s Next
With youth adoption of AI mental health tools continuing to climb and state-level restrictions expanding, 2026 looks likely to be a turning point year for how the industry, regulators and clinicians negotiate the role AI should play in mental health support. Researchers say the central unresolved question is not whether young people will keep using these tools, since that trend appears well underway, but whether the technology and its guardrails can be made trustworthy enough to match the scale at which it is already being used.
What Parents and Schools Are Doing
School counselors and pediatricians surveyed about the RAND findings say they are increasingly incorporating conversations about AI chatbot use directly into standard mental health screenings, treating it as a routine topic similar to asking about social media use. Some school districts have begun developing guidance for teachers and counselors on how to talk with students who disclose reliance on AI chatbots for emotional support, while pediatric associations have called for more research into which specific chatbot use patterns, occasional venting versus primary reliance during a genuine crisis, carry meaningfully different risk profiles for young users.
