Authored By Jan Warner

This article examines selected ethical issues in AI-mediated psychotherapy, focusing on measuring its clinical efficacy and noting the limited evidence available for specific psychological disorders and populations. Current AI therapy research often relies on established measures of symptom reductions that were not designed specifically to measure AI psychotherapy and may overlook characteristics unique to AI-mediated treatment. This lack of validated AI-specific symptom outcome measures complicates the evaluation of research and commercial claims. Additional concerns include inconsistent definitions of AI interventions, limited long-term follow-up measures, cultural generalizability, digital literacy, underrepresentation of diagnostic and demographic groups, and the potential role of AI in assessment and diagnosis. Greater interdisciplinary collaboration, conceptual precision, culturally responsive research, standardization of terminology, validated methods of evaluating therapeutic processes and clinical outcomes are needed before AI interventions can be considered an ethical mental health care practice.