In an English school survey, 610 of 662 young people using AI for mental-health support also reported other help. The 92.1% overlap places chatbot support within a wider network.[1]
The wider 2025 OxWell project included 35,244 participants across primary schools, secondary schools and further education colleges. Most responses were collected from 24 February to 28 March, with six schools continuing in April and May.[2]
The AI analysis, published on 27 October, covers 24,132 secondary and college respondents, mainly in Oxfordshire and Liverpool. Recruitment was not a national probability sample.[1]
Among the AI-support users, 305 reported formal services. Separately, 324 of 580 respondents rated the AI support helpful, while 410 of 649 wanted additional support. Different denominators reflect different questions answered. Helpfulness and unmet need can coexist.[1]
Why the question changes the number
Two estimates measuring different kinds of use
| Study | What was counted | Reported share |
|---|---|---|
| OxWell | Past-year AI/LLM use for mental-health support, among 24,132 respondents. | 2.7%Source 1 |
| JED | AI use when sad, stressed, lonely or overwhelmed, among 5,560 students at 21 US schools. | 57%Source 3Source 4 |
JED collected its responses between October 2025 and April 2026. Its announcement explicitly distinguishes using AI while struggling emotionally from recognising that activity as getting help. That framing matters when placing its estimate beside a survey organised around sources of mental-health support.[3][4]
Consider an illustrative student who asks a chatbot to distract them after an argument. They might report using AI when upset while declining to call it mental-health support. Both answers could accurately describe the same activity. A percentage becomes interpretable only when its question travels with it.
The survey measures reported use and helpfulness. Having AI and other help during the same year establishes neither treatment effectiveness nor which source came first.[1]
For the next survey, a valuable question is what happens between those sources: whether a conversation leads someone towards another person, fills a waiting period, or becomes the only place a particular concern is voiced. Counting chatbot users alone cannot distinguish those routes.
Sources & further reading
- General-purpose AI as a source of mental health support amongst 24,000 adolescents in England npj Mental Health Research · 27 Oct 2026 · Accessed 28 Oct 2026
- The OxWell Student Survey 2025 – Study Description University of Oxford Department of Psychiatry · 30 Mar 2026 · Accessed 28 Oct 2026
- Exploring Artificial Intelligence: Understanding Relationships between AI Use and Mental Health Among Middle and High School Students The Jed Foundation · 27 Oct 2026 · Accessed 28 Oct 2026
- New JED Research Shows Link Between Teens’ AI Use, Suicidal Thoughts and Behaviors The Jed Foundation · 27 Oct 2026 · Accessed 28 Oct 2026