Generative AI in Youth Mental Health Apps: Rapid Review

Key Findings

• — Only five studies met the inclusion criteria, and the most common integration route was a chatbot inside the app
• — Young people generally rated these apps well on usability and quality, and some studies reported promising effects on depression, anxiety and distress
• — None of the included studies systematically assessed harmful effects using standardised methods, and none reported any adverse mental health outcomes
• — The authors state that the absence of reported adverse outcomes may reflect the absence of investigation rather than evidence that no harm occurred
• — The recommendation is for research that maps possible adverse events systematically alongside short- and long-term effects

Methodology Notes

Rapid review methodology, four databases searched November 2025, inclusion limited to primary research on samples aged 11 to 24 published 2022 to 2025 in English or a Scandinavian language. The small yield of five studies is itself the substantive result and also the review's main limitation: conclusions about effect rest on a very thin evidence base, and the language restriction excludes non-European literature. Author affiliation is the Regional Centre for Child and Youth Mental Health and Child Welfare, Northern Norway, at UiT The Arctic University of Norway. Verification route: mental.jmir.org returned an empty response to this host, so the record was verified against the Europe PMC core record (PMID 42615471, full abstract read) and the Crossref deposit (DOI 10.2196/90589, published 2026-08-19, JMIR Mental Health), which agree on title, authors, venue and date.

Sources