Contextualizing the privacy paradox—a risk–benefit analysis of generation z’s adoption intentions toward AI-based virtual try-on

Introduction: With the rapid advancement of artificial intelligence (AI), AI-driven virtual try-on (AI-VTO) services are reshaping consumption patterns in fashion retail. At the same time, their reliance on sensitive personal data has intensified privacy-related concerns. As digital natives and a key consumer segment, Generation Z often exhibits a "privacy paradox" in AI-enabled contexts, expressing concern about privacy while continuing to use data-intensive services. To explain this phenomenon, this study integrates the Theory of Planned Behavior (TPB) and the Privacy Calculus Model (PCM) into a unified risk-benefit framework. Methods: Survey data were collected from 709 Generation Z consumers in the Yangtze River Delta region of China. The data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) and Fuzzy-set Qualitative Comparative Analysis (fsQCA). Results: The results show that perceived responsiveness, attitude, and perceived behavioral control positively influence intention to use AI-VTO services, whereas intrusiveness concerns exert a significant negative effect. In contrast, traditional privacy concerns do not have a direct effect on usage intention. Attitude mediates the effects of both perceived benefits and perceived risks on behavioral intention. In addition, the fsQCA results identify three distinct pathways leading to high adoption intention: an efficacy trust-driven pathway, an experience-driven pathway, and a control convenience-driven pathway. These findings suggest that the privacy paradox is more likely to emerge in experience-oriented contexts. Discussion: This study clarifies how Generation Z evaluates data-intensive AI services by revealing both net effects and configurational pathways underlying AI-VTO adoption. It extends current understanding of the privacy paradox in AI-enabled consumption and offers practical implications for developing transparent, user-centered, and trustworthy AI-VTO systems.