A Mass General Brigham study analyzing the 2024-2025 Healthy Minds Study found that 18% of 675 surveyed college students used generative AI for mental health support, per findings published in the Journal of Affective Disorders. Students with moderate depression, severe depression, severe anxiety, or suicidality were each about two-fold more likely to use AI for mental health, and Asian students had roughly twice the odds compared with peers, according to the study. Lead author Cindy H. Liu, PhD (Mass General Brigham), is quoted: "College students who are most drawn to AI for mental health may also be the most vulnerable to its risks." In a second quote Liu adds: "AI can act as a relational partner that is always available, never rejects, and offers unconditional validation," noting that empirical evidence on harms or benefits remains limited. The investigators recommend that AI platforms embed crisis detection and referral mechanisms, and that institutions track and respond to student AI use alongside formal care.
What happened
A peer-reviewed study led by Cindy H. Liu, PhD, director of the Developmental Risk and Cultural Resilience Laboratory at Mass General Brigham, analyzed data from the 2024-2025 Healthy Minds Study, an annual web-based survey of U.S. college student mental health. Among 675 students from two institutions, 18% reported using generative AI for mental health, per findings published in the Journal of Affective Disorders (DOI: 10.1016/j.jad.2026.122058) and reported by Mass General Brigham and EurekAlert.
Key findings
Four separate symptom categories - moderate depression, severe depression, severe anxiety, and suicidality - were each associated with an approximately two-fold higher likelihood of AI use for mental health, per the study. Asian students also had about twice the odds of using AI for mental health compared with peers, according to Liu et al.
Researcher quotes
Liu is quoted in the Mass General Brigham press release: "College students who are most drawn to AI for mental health may also be the most vulnerable to its risks. College students who are struggling may seek out AI, and we worry that these unregulated tools could stand in for human support. At the same time, many students clearly find these tools useful, which is a reason to understand where they help and where they fall short."
In a second quote Liu adds: "Conversations with AI for mental health may pose a risk because of how appealing they are: AI can act as a relational partner that is always available, never rejects, and offers unconditional validation. We don't yet know whether using general-purpose AI for mental health is beneficial or whether it undermines critical capacities such as emotional regulation or perspective-taking."
Recommendations
The investigators provide actionable guidance for three stakeholders, per the paper: AI platforms should embed crisis detection and referral mechanisms; institutions should consider how to support students who turn to AI when formal care feels inaccessible - a pattern observed among students with severe depression and Asian students; and mental health practices should seek to understand how patients are using these tools alongside or in place of formal care.
Limitations and context
The sample of 675 students from two institutions limits generalizability, and the cross-sectional design of the Healthy Minds Study means the associations are correlational. Observed industry patterns in comparable deployments show a regulatory gap: mainstream conversational AI systems are not required to include clinically validated triage, crisis detection, or human-escalation pathways. The study provides a data point on who is using these tools but not on outcomes, a gap the investigators explicitly flag as requiring further research.
What to watch
Whether larger, representative Healthy Minds Study releases replicate the two-fold association; whether AI vendors adopt embedded crisis-detection and referral features; and any regulatory guidance addressing conversational AI safety in mental health contexts.
Scoring Rationale #
Peer-reviewed finding in the Journal of Affective Disorders documenting higher generative AI adoption among vulnerable college students, with actionable safety recommendations for AI developers and campus health institutions. Relevant to responsible AI deployment in mental health contexts, but generalizability is constrained by the small sample of 675 students from two institutions; does not change core AI capabilities or policy.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems