Background AI-driven music therapy offers a promising, accessible digital intervention for the growing mental health crisis in universities. The “Deficiency Compensation Hypothesis” suggests that depression may drive students toward such digital help-seeking. However, the inherent data sensitivity of AI tools triggers the “Privacy Calculus,” potentially inhibiting adoption. This study investigates the interplay between depression severity, privacy concerns, and the intention to use AI music therapy among university students. Methods A cross-sectional survey was conducted with 612 university students in China. The study measured depression levels (PHQ-8), AI-specific privacy concerns, perceived usefulness, and intention to use. A hierarchical regression model with moderation analysis was employed to examine whether privacy concerns weaken the association between distress and help-seeking motivation. Results Participants exhibited mild depression on average (PHQ-8 Mean = 6.07). Regression analysis revealed that depression positively predicted the intention to use AI music therapy ( β = 0.128, p 0.001 ), supporting the distress-driven help-seeking hypothesis. Crucially, privacy concerns acted as a significant negative moderator ( β = −0.086, p = 0.015 ). Simple slope analysis indicated that the motivating effect of depression on usage intention was significant only for students with low privacy concerns but was nullified in those with high privacy concerns. Conclusion The findings highlight a critical paradox in digital mental health: while depressive symptoms are positively associated with students’ intention to seek AI-based help, privacy fears can significantly attenuate this association. For highly privacy-sensitive individuals, the need for therapeutic relief is overridden by the fear of surveillance. Consequently, developers and universities must prioritize “privacy by design” and transparent trust mechanisms, rather than relying solely on algorithmic precision, to ensure these tools can serve as effective emotional support for vulnerable students.
Zhu et al. (Tue,) studied this question.