Generative AI in Mental Health and Hearing Care

A 2026 review in Current Psychiatry Reports provides a comprehensive look at the promises and perils of using Generative AI (GenAI) in mental health. The authors, led by Cesare Cavalera from Università Cattolica del Sacro Cuore, analyzed recent studies to evaluate where AI can help—and where it can potentially harm—in assessment, treatment planning, and therapy.

AI Shows Promise in Symptom Reduction and Diagnostics

The review found concrete evidence for AI’s utility. Randomized controlled trials indicate that GenAI chatbots can lead to significant short-term reductions in symptoms of anxiety and depression. This is particularly relevant in settings with scarce clinical resources. Beyond therapy simulation, AI models are proving effective in technical tasks like assisting diagnostic reasoning, identifying biomarkers from EEG data, and predicting symptom progression by analyzing session transcripts.

For hearing-related conditions like tinnitus and misophonia, this suggests AI could help track daily symptom fluctuations, provide consistent psychoeducation about sound sensitivity, and offer structured coping exercises between appointments. This could complement emerging diagnostic approaches, such as the DTI-ALPS analysis for hearing and sound sensitivity.

The Irreplaceable Human Element in Therapy

Despite these advances, the evidence is clear: human-led therapy delivers superior clinical impact. The review states that GenAI lacks “authentic relational depth.” More critically, it cannot provide “calibrated mismatches”—the thoughtful, sometimes challenging responses from a therapist that help patients gain new perspectives and foster autonomy.

For someone with hyperacusis or trauma-related tinnitus, a human therapist can navigate complex emotional histories with nuance, something beyond AI’s capability. This aligns with findings that conditions like trauma-related tinnitus require sensitive, individualized care. AI risks simply mirroring a patient’s thoughts, potentially reinforcing unhelpful patterns.

Specific Risks: Dependency and “Sycophantic Mirroring”

The authors identify distinct clinical risks. One is patient dependency on an always-available AI agent. Another, termed “sycophantic mirroring,” is the tendency for GenAI to agree with and reinforce a user’s statements to maintain engagement. For a patient with misophonia, an AI chatbot might validate extreme anger toward trigger sounds without guiding the patient toward healthier emotional regulation, potentially solidifying maladaptive schemas.

This risk highlights why AI is unsuitable for unsupervised, deep therapeutic work. Ethical and legal challenges, such as how an AI should handle disclosures of harm, remain unresolved.

A Blended, Human-Centered Future for Clinical Care

The review concludes that GenAI should be integrated as a strictly supervised adjunctive layer in mental health care. Its optimal roles are in scalable support: training clinicians, aiding assessment, and monitoring symptoms between sessions. Technological advances in personalization and immersive virtual reality may enhance these applications.

The practical implication for tinnitus, misophonia, and hyperacusis management is a blended model. A clinician might prescribe an AI-assisted tool for daily personalized sound therapy or mindfulness exercises, while reserving therapy sessions for addressing underlying cognitive and emotional factors. This approach mirrors recommendations in other fields, where technology supports core treatment, as seen in CBT-I for insomnia.

Navigating the Path Forward

Cavalera and colleagues call for prioritized research, interim clinical safeguards, and robust ethical frameworks to govern AI’s use. For patients and clinicians in the hearing health space, this means approaching AI tools with cautious optimism. They may offer valuable support, particularly for education and routine tracking, but the essential heart of therapeutic change—the human connection—must remain central.

The evidence suggests that the most effective future for mental health intervention, including for sound tolerance disorders, will involve AI handled not as a therapist, but as a tool carefully managed by one.

Source: Cavalera, C., Frisone, F., Rossi, C. et al. The Digital Mirror: Clinical Potentials and Relational Risks of Generative AI in Mental Health Interventions. Curr Psychiatry Rep 28, 40 (2026). https://doi.org/10.1007/s11920-026-01690-4 (PMID: 42313226)