The NSF AI Research Institute on Interaction for AI Assistants (ARIA) submitted this comment to the FDA on October 19, 2026, recommending a risk framework for generative AI-enabled medical devices that accounts for how people use and rely on AI over time, not just the quality of individual responses. The comment focuses on patient-facing conversational AI for mental and behavioral health.

ARIA’s central finding is that risks can accumulate within a single conversation and across repeated use. Chatbots may amplify concerns, reinforce distorted beliefs, contribute to dependence, or change whether people seek support from clinicians, friends, and family. The FDA’s assessment of safety and effectiveness should therefore reflect realistic patterns of use, including frequent and prolonged use.

ARIA is a National Science Foundation AI Institute that brings together more than 40 faculty and 50 students, postdocs, and researchers at 12 institutions, including Brown University, Carnegie Mellon University, Colby College, Dartmouth College, Data & Society, New York University, Princeton University, Santa Fe Institute, UC Berkeley, UC San Diego, the University of New Mexico, and the University of Virginia.

The comment makes five recommendations, which you can read here.