College Students' Trust in Generative AI for Mental Health.

OBJECTIVE

This study aimed to examine college students' trust in generative artificial intelligence (AI) for mental health information and decisions.

METHODS

Among 926 undergraduates at two institutions in the 2024-2025 Healthy Minds Study, demographic and clinical correlates of dichotomized trust ratings were examined.

RESULTS

Overall, 31% trusted AI for mental health information, and 13% trusted AI for mental health decisions. Severe depression, anxiety, and therapy history were not significantly associated with trust. Students with (vs. without) a lifetime psychiatric diagnosis trusted AI less for information (28% vs. 34%). Lesbian, gay, bisexual, and queer (LGBQ; vs. non-LGBQ) students trusted AI less for information (21% vs. 38%) and decisions (7% vs. 17%). International (vs. domestic) students trusted AI more for information (52% vs. 29%), and Asian (vs. White) students trusted AI more for decisions (17% vs. 11%).

CONCLUSIONS

Trust varied by identity but not by most clinical characteristics. Clinicians should ask patients about AI use.