TAAC: Trust Framework for Audio-Based AI Depression Diagnosis Identifies Security Risks in Clinical Deployment

arXiv·low signal

Examines security and trust vulnerabilities in audio-based AI depression diagnosis systems gaining clinical adoption. Identifies attack surfaces specific to affective computing: adversarial audio perturbations that flip diagnostic outcomes, privacy leakage from voice biomarkers, and model manipulation risks in clinical settings. Proposes the TAAC trust framework for securing audio-based mental health AI as these systems move from research prototypes to production healthcare deployment.