That represents a 60% increase in one year, turning AI from a private experiment into a meaningful part of how many young people try to find support. For patients, families, and clinicians, the immediate question is no longer whether AI is being used, but how its advice is being evaluated.
A private route into care
The Harvard survey included 1,727 participants and found that AI use varied across youth groups: relatively older young people were more likely to use it, as were females. Most young people who used AI did not tell an adult, including a parent or physician.
That silence matters because general-use AI models are not designed to provide mental-health advice, and researchers say misinformation is available across different systems. A chatbot can sound calm, attentive, and confident even when the underlying guidance is incomplete or wrong. The emotional experience of being heard can therefore become difficult to separate from the reliability of what is being said.
At the same time, the appeal is easy to understand. AI is available around the clock, does not require an appointment, and is unlikely to sound impatient or harsh. Those features may be especially attractive to people living in rural or remote areas, where mental-health professionals are in short supply. Harvard researchers also pointed to an acute shortage of psychiatrists, particularly child psychiatrists, and said only half of people with mental-health needs receive treatment.
This is the central paradox: the qualities that make AI feel accessible may also make it easier to trust without scrutiny.
Demand is rising beyond the chatbot
The broader need for youth support is also visible in data from Canada. Kids Help Phone recorded a 24% increase in overnight text conversations during the first half of 2026 compared with the same period a year earlier. The service receives, on average, a text every 90 seconds.
These figures do not establish that AI use is causing the wider increase in demand, and they come from different countries and sources. Together, however, they describe a landscape in which young people are seeking help at times and in ways that traditional services may not always reach. AI can fill a gap in availability, but availability alone does not establish safety or clinical quality.
The Harvard researchers said harmful advice remains a concern and noted that tragedies, including suicide attempts, have occurred after young people discussed mental-health problems with AI. That does not mean every conversation with a chatbot is harmful; it does mean we should avoid treating a fluent response as a clinical assessment or a dependable care plan.
What to anchor to before relying on AI
If you or someone close to you is using a chatbot for mental-health concerns, the most useful first step is not shame or a sudden demand to stop. It is to make the use visible to a trusted adult, physician, or mental-health professional. Knowing what was asked and what the system advised gives another person a chance to help distinguish accurate information from misinformation.
We can also hold two ideas at once: a chatbot may offer immediate companionship or help someone put difficult feelings into words, while still being an unreliable source for decisions about mental health. A small grounding habit is to save the response that feels important, wait before acting on it, and bring it into a conversation with a qualified professional or trusted adult. That pause helps us separate the comfort of being answered from the harder question of whether the advice is sound.