Trustworthy healthcare AI for wellbeing: how AI characteristics and information sensitivity shape privacy disclosure

As artificial intelligence (AI) becomes increasingly integrated into healthcare, understanding how AI characteristics and data sensitivity influence users’ disclosure behaviour is critical for designing trustworthy and emotionally safe AI systems. This study investigates how two types of AI trust cues (i.e. technical vs. psychological securities) and the type of personal information requested (i.e. demographic vs. digital footprint) interact to shape perceived trust, disclosure intention, and actual disclosure behaviour in the course of AI interaction. We conducted a 2 × 2 between-subjects online experiment with 146 participants using a Wizard-of-Oz healthcare chatbot. Results showed a significant interaction effect on disclosure intention. Participants showed higher intention to disclose demographic information to psychologically secure AI, while technical security framing increased disclosure intention for digital footprint information. Trust significantly mediated these effects. Although AI characteristics did not affect the amount of disclosed data, psychological security led to significantly deeper, more emotionally intimate disclosures. These findings extend trust and privacy research by revealing that AI trust cues operate differently depending on information sensitivity. The interaction demonstrates that effective trust-building strategies must be tailored to the specific disclosure context.