Expert or Partner? How Social Roles of AI Chatbots Shape Competence Trust and Privacy Disclosure Intention in Fitness Contexts

In fitness contexts, the use of conversational artificial intelligence (AI) chatbots has made the disclosure of highly sensitive information a prerequisite for the effective operation of AI fitness services. However, existing research has rarely explored how the social roles presented by AI influence user trust and privacy decisions across varying contexts of high-sensitivity information. Drawing upon Social Role Theory and the perspective of competence trust, this study focuses on fitness AI chatbots and employs a 2 (Social Roles: Expert vs. Partner) × 2 (High-Sensitivity Information: Personally Identifiable Information vs. Fitness Data) between-subjects experimental design to systematically examine the role-information congruence effect, as well as the underlying psychological mechanisms. Based on an empirical analysis of 686 valid samples, the results indicate that social roles and high-sensitivity information exert a significant interaction effect on users’ competence trust, an effect primarily observed among risk-averse users. Furthermore, competence trust has a significant positive influence on privacy disclosure intention, while perceived risk and perceived benefit do not exert stable moderating effects. The study further reveals that risk preference primarily functions as an antecedent individual difference variable in the formation of competence trust. This research extends the contextual understanding of AI anthropomorphism and privacy disclosure within fitness scenarios and provides empirical evidence for social role design and trust management in highly sensitive, data-driven AI services.