Artificial intelligence is increasingly being used to support mental health, presenting both promising opportunities and significant risks that experts say require careful navigation. According to Harvard Business Review, AI-powered therapy and companionship will dominate generative AI use cases in 2025, raising questions about whether this represents a mental health revolution or uncharted dangers.
The Case for AI Support
The need for AI-assisted mental health interventions is urgent: nearly 1 in 4 U.S. adults (58+ million people) have a mental health condition, while 1 in 3 live in areas with therapist shortages, according to researchers cited in a 2024 paper. Nicholas C. Jacobson, PhD, associate professor of biomedical data science and psychiatry at Dartmouth’s Geisel School of Medicine, states that “yes, if used responsibly,” AI can help. Eduardo Bunge, PhD, associate director for the International Institute of Internet Interventions for Health at Palo Alto University, notes that “if you don’t have access to a therapist, AI is better than nothing” because it provides immediate access when people are struggling most.
Another benefit is that people often share more with AI than with human therapists, even without fear of judgment. In one small study, 80% of participants found ChatGPT helpful for managing symptoms. However, Jacobson cautions that some users complained about AI’s “guardrails,” which shut down conversations about issues requiring higher-level care, like suicide.
Critical Risks to Know
The dangers are substantial. Jacobson warns that “AI almost always sounds incredibly fluent, even when it’s giving you harmful responses,” and this convincing tone doesn’t guarantee quality or evidence-based care. Alexandra Werntz, PhD, associate director of UMass Boston’s Center for Evidence-Based Mentoring, emphasizes that “AI isn’t a human who is looking out for you” and that “responses can be incorrect, culturally biased, or harmful.” Privacy is another concern—these tools are not HIPAA-compliant, and users should not share information they wouldn’t want others reading.
Safe Uses and Future Directions
AI works best as a supplement to therapy, not a replacement. Experts recommend using it to identify potential symptoms, explore evidence-based treatments, understand different therapy types, reframe negative thoughts, and practice social skills. However, AI should never be used for suicidal thoughts, self-harm urges, psychosis symptoms, or eating disorders, where human professionals are essential.
Looking ahead, therapy bots specifically programmed for mental health show promise. In March 2025, Dartmouth researchers published the first randomized controlled trial showing that Therabot significantly reduced symptoms in people with major depressive disorder, generalized anxiety disorder, and those at high risk for eating disorders. Jacobson remains optimistic that AI represents “a huge game changer for the availability of high-quality mental health care,” though human therapists will remain essential.
This article is an AI-assisted summary. All facts and figures are drawn from the original report: https://www.newsweek.com/how-ai-can-help-or-hurt-your-mental-health-5-prompts-experts-recommend-11703293