What question did this study set out to answer?

The review aims to assess the clinical applications and risks of generative AI in mental health care.

June 20, 2026Open Access

The Digital Mirror: Clinical Potentials and Relational Risks of Generative AI in Mental Health Interventions

CCCesare CavaleraUniversità Cattolica del Sacro Cuore FFFabio FrisoneUniversità Cattolica del Sacro Cuore CRChiara RossiUniversità Cattolica del Sacro Cuore

Key Points

The review aims to assess the clinical applications and risks of generative AI in mental health care.
Reviewed recent studies and randomized controlled trials on generative AI in mental health interventions
Analyzed AI applications in assessment, treatment planning, and psychotherapy
Critically examined ethical dilemmas and clinical risks associated with AI use
Randomized controlled trials indicate GenAI chatbots significantly reduce anxiety and depression symptoms
Human-led therapy shows greater emotional engagement and clinical impact compared to GenAI
Identified risks include potential dependency and ethical-legal challenges from AI engagement

Abstract

PURPOSE OF REVIEW: This review explores the rapidly evolving integration of Generative Artificial Intelligence (GenAI) in mental health care. It aims to evaluate current applications in assessment, treatment planning, and psychotherapeutic interventions, while critically examining the clinical risks, ethical dilemmas, and the future potential of GenAI as an adjunctive tool rather than a replacement for human-delivered therapy. RECENT FINDINGS: Recent studies indicate that AI models can effectively assist in diagnostic reasoning, biomarker identification via EEG, and the prediction of symptom trajectories from session transcripts. Randomized controlled trials (RCTs) suggest that GenAI chatbots significantly reduce anxiety and depressive symptoms in the short term, particularly in settings with limited access to clinicians. However, human-led therapy remains superior in fostering deep emotional engagement and clinical impact. Significant risks identified include the potential for GenAI to foster dependency, reinforce maladaptive schemas or delusional ideation through "sycophantic" mirroring, and raise complex ethical-legal challenges regarding the reporting of criminal disclosures. AI represents a transformative adjunctive layer in mental health, offering scalable support for assessment, training, and between-session monitoring. While technological advances in personalization, multimodality, and immersive virtual reality enhance its clinical utility, GenAI lacks the authentic relational depth and"calibrated mismatches" essential for autonomy and transformative change. Future integration must prioritize a human-centered, blended approach, where GenAI is strictly supervised by clinicians within a robust ethical and regulatory framework to preserve the essential heart of the therapeutic connection. Research priorities, interim clinical safeguards, and recommendations for navigating the gap between current evidence and real-world adoption need to be defined and implemented.

Ask AI

Helpful

Bookmark

View Full Paper