AI Therapy Chatbot Delivers Results in Depression and Anxiety Treatment
A clinical trial shows that AI-powered chatbot, Therabot, reduced symptoms in depression, anxiety and eating disorders.

Complete the form below to unlock access to ALL audio articles.
A clinical trial conducted by researchers at Dartmouth College has shown that a generative AI chatbot, developed as a mental health support tool, significantly reduced symptoms in individuals with depression, anxiety or eating disorders. The results were published March 27 in New England Journal of Medicine AI.
The trial, which involved 106 participants across the United States, tested a software tool called Therabot. Participants diagnosed with major depressive disorder, generalized anxiety disorder or eating disorders used a smartphone app to engage in text-based interactions with the chatbot. The conversations were either prompted by the app or initiated by the user.
Major depressive disorder
A common and serious mood disorder characterized by persistent sadness, loss of interest or pleasure, and a range of physical and emotional problems that interfere with daily functioning.
Generative AI
A type of artificial intelligence that can generate human-like text, images or other outputs. In mental health applications, it may be used to simulate conversation or deliver therapeutic interventions.
Generalized anxiety disorder
A mental health condition involving persistent and excessive worry about various aspects of life, often accompanied by restlessness, fatigue, difficulty concentrating and sleep disturbances.
Clinical improvements observed in multiple mental health conditions
Over a four-week period, users with depression reported an average 51% reduction in symptoms, while those with generalized anxiety showed a 31% reduction. Among users at risk for eating disorders, concerns related to body image and weight declined by 19%. These improvements were measured using standardized questionnaires and compared to a control group of 104 individuals with the same conditions who did not use the app.
Want more breaking news?
Subscribe to Technology Networks’ daily newsletter, delivering breaking science news straight to your inbox every day.
Subscribe for FREEParticipants in the Therabot group were not required to be undergoing other forms of treatment, and nearly three-quarters were not receiving medication or therapy at the time. The app adjusted its questions and responses based on users' input, aligning with cognitive behavioral therapy and other evidence-based approaches.
Cognitive behavioral therapy
A type of psychotherapy that helps individuals identify and change negative thought patterns and behaviors. It is commonly used to treat depression, anxiety disorders and other mental health conditions.Chatbot engagement reflects trust and therapeutic alliance
According to the study, users spent an average of six hours interacting with Therabot over the four-week intervention – roughly the equivalent of eight therapy sessions. Importantly, the software maintained a strong "therapeutic alliance" with users, a term that refers to the sense of trust and collaboration between patient and caregiver. This metric is typically linked to better treatment outcomes in clinical settings.
Therapeutic alliance
The collaborative and trusting relationship between a patient and a mental health professional, which is considered critical for successful therapeutic outcomes.The chatbot also allowed users to engage during off-hours, with increased activity observed during late-night hours, a time often associated with psychological distress.
Safety protocols and clinician oversight remain essential
Though the study showed encouraging results, researchers emphasized that generative AI in mental health care still requires careful oversight. The chatbot was programmed to detect high-risk content such as suicidal ideation, at which point it prompted users to seek immediate help through emergency services or crisis lines. Conversations were monitored to ensure responses adhered to therapeutic best practices.
Therabot has been under development since 2019 by Dartmouth’s AI and Mental Health Lab. Throughout its development, psychologists and psychiatrists were consulted to guide the chatbot’s content and response patterns. Previous evaluations of the chatbot indicated that over 90% of its outputs aligned with established therapeutic standards.
Potential for scalability, but with caution
The study’s authors note that while AI-based tools cannot replace traditional mental health care, they may help fill gaps in access. With demand far outstripping the availability of licensed mental health professionals, AI-powered applications could offer scalable support options, especially for those unable to engage in regular therapy sessions.
Researchers also stress that AI tools must be designed with mechanisms to manage and respond to sensitive or high-risk disclosures. While no major safety concerns were observed in this trial, the unpredictability of generative models highlights the importance of continued oversight and ethical safeguards in digital mental health care.
Reference: Heinz MV, Mackin DM, Trudeau BM, et al. Randomized trial of a generative AI chatbot for mental health treatment. NEJM AI. 2025;2(4). doi: 10.1056/AIoa2400802
This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source. Our press release publishing policy can be accessed here.
This content includes text that has been generated with the assistance of AI. Technology Networks' AI policy can be found here.