AI Therapy Chatbots Showed Clinical Potential

New research suggests AI tools may help manage mental health, though regulation debates continue in New Hampshire.

Updated on Sept. 22, 2026 in Mental Health

Bold vector editorial illustration of a glowing sphere on a pedestal, representing digital wellness support in a clean, uncluttered space.
New research in New Hampshire indicates that AI-driven therapy chatbots like Therabot and Evergreen can effectively reduce symptoms of anxiety and depression. AI Illustration. Upload story photo >

Live Poll

Would you feel comfortable using an AI chatbot for mental health support without human clinician oversight?

AI-driven therapy tools known as Therabot and Evergreen have demonstrated potential for improving mental health outcomes, including symptom reduction for anxiety and depression. These digital interventions remain under active study by researchers as state-level conversations regarding their regulation persist.

Why it matters

The development of these tools aims to expand access to mental health support through AI-based interfaces. Ongoing research seeks to balance innovation with clinical oversight, especially as legislative efforts to mandate manual clinician review for every exchange were recently rejected.

The first clinical trial of Therabot, which concluded in March 2025, demonstrated symptom improvement for anxiety, eating disorders, and depression. This preliminary evidence marks the initial evaluation of the generative AI platform within a clinical study framework.

The players

Nicholas Jacobson

A professor of psychiatry at Dartmouth College who leads the development and clinical testing of AI-based mental health interventions.

Michael Heinz

A researcher and collaborator with Nicholas Jacobson who co-developed the Therabot platform for mental health support.

New Hampshire House

The legislative body that recently rejected Senate Bill 640, a proposal concerning the regulatory framework for therapy chatbots.

The details

Therabot functions as an intervention utilizing generative AI through text chat, voice modes, and an animated avatar to simulate therapeutic engagement. The companion tool, Evergreen, is designed to support flourishing and goal-seeking behaviors by integrating with optional student data from platforms like Canvas to personalize the experience.

Timeline

  1. Therabot development began in 2019.

  2. The first clinical trial for Therabot concluded on March 27, 2025.

  3. Researchers published an op-ed opposing Senate Bill 640 in January 2026.

  4. Additional trials for the Evergreen chatbot are planned for the fall of 2026.

  5. Trials for the generative version of Evergreen are scheduled to begin in 2027.

Health Landscape

The rejection of New Hampshire Senate Bill 640 highlights the current regulatory tension regarding whether digital health AI requires manual clinician oversight for every patient interaction. This event marks a departure from stricter regulatory requirements that would have mandated manual review.

As these AI tools emerge, patients should consider whether digital interfaces serve as an appropriate complement to their existing care plans. Always consult with your doctor or a licensed mental health professional before relying on AI chatbots for symptom management or diagnostic concerns.

The takeaway

While AI therapy tools are evolving, they currently function as emerging technology rather than established clinical practice. If you are exploring digital mental health resources, discuss the specific platform's level of professional oversight with your provider to ensure it fits your needs.

What happens next

Evergreen chatbot trials are scheduled for the fall of 2026, followed by trials for the generative version of the platform in 2027.

Further reading

For more on the intersection of technology and clinical care, see our Mental Health section.

Source note: This article includes information reported by The Dartmouth.

Live Poll

Would you feel comfortable using an AI chatbot for mental health support without human clinician oversight?