In a real-world pilot, Ash users saw sustained improvements in depression and anxiety

A 2025 single-arm study of Ash users found sustained reductions in depression and anxiety and gains in social support, with safety guardrails handling all flagged sessions.

Among adults using Ash over a 10-week period, depression (PHQ-9) and anxiety (GAD-7) symptoms fell significantly and held at follow-up, alongside gains in hope, social connection, and perceived social support.

Study type

Single-arm naturalistic observational study

Published in

arXiv

Partner

New York University

Citation

Hull, T. D., Zhang, L., Areán, P. A., & Malgaroli, M. (2025). Generative AI Purpose-built for Social and Mental Health: A Real-World Pilot [Preprint]. arXiv. https://arxiv.org/abs/2511.11689

A 2025 arXiv preprint reported a single-arm naturalistic study of adults using Ash, a purpose-built generative-AI tool for mental health, between May and September 2025. Participants showed significant, sustained reductions in depression and anxiety and improvements in hope, behavioral activation, loneliness, and perceived social support. Engagement and working alliance — comparable to traditional care — predicted outcomes, and automated safety guardrails handled all 76 risk-flagged sessions per escalation policy.

Improvements in PHQ-9 and GAD-7 were sustained at the 10-week follow-up, and gains in hope, behavioral activation, social interaction, loneliness, and perceived social support were maintained throughout. Higher engagement and a stronger working alliance both predicted better outcomes.

As a single-arm observational study without a control group, it cannot establish causation; the authors frame it as initial real-world evidence that builds on earlier randomized designs. Co-authors include external researchers Patricia A. Areán and Matteo Malgaroli.