Patients are increasingly using generative AI tools, such as ChatGPT, to cope with distress, self-diagnose, or seek connection and support. While some uses may be adaptive, such as practicing skills between sessions or accessing psychoeducation, others may inadvertently heighten risk by reinforcing maladaptive coping, spreading misinformation, or providing content that is unsafe. Clinicians are now encountering patients who arrive with AI-driven self-diagnoses, treatment information, or use of conversational agents for emotional regulation. This presents a new clinical imperative: how to assess, contextualize, and guide patient engagement with AI in ways that safeguard well-being and preserve treatment integrity.
This roundtable will address how clinicians can integrate AI assessment into intake and ongoing care. Panelists will outline structured approaches for asking about AI use in session, and will discuss frameworks for differentiating potentially adaptive engagement (e.g., journaling prompts, guided exposure practice, psychoeducation) from potentially harmful patterns (e.g., compulsive reassurance-seeking or use of unregulated tools for crisis management). Panelists will also discuss different AI tools and how they differ in terms of clinical safeguards. Finally, the conversation will explore strategies for assessing AI use within populations that may be more vulnerable, such as children and individuals with serious mental illness.
Ethical and practical considerations will be central. Topics will include how to assess for and respond when AI-generated content contradicts evidence-based practice, how to document and address risky use, and how to communicate guidance without alienating patients who value these tools. We will also examine broader systemic concerns, including bias within algorithms, the lack of regulation, and the potential for disparities in who benefits or is harmed by AI.
Participants will leave with strategies for incorporating assessment of AI use into their clinical formulation and supervision practices, as well as language for discussing AI engagement in ways that are collaborative and clinically productive. This roundtable is intended for clinicians at all levels who seek to adapt their practice for a digital landscape.
@inproceedings{sarma2026adaa,
author = {Berman, G. and Nurmi, E. and Sarma, K. V. and Zaboski, B. A.},
title = {The AI SIG Presents: An Unregulated Innovation: Guiding Patient Use of Generative AI in Anxiety and Depression Treatment},
booktitle = {ADAA Annual Meeting},
year = {2026},
}