Does empathetic artificial intelligence influence users’ intention to accept health misinformation?

With the rapid adoption of generative artificial intelligence (AI) featuring empathetic expressions in health consultation contexts, concerns have emerged regarding the risk of health misinformation dissemination alongside improved interaction quality. Grounded in the stimulus–organism–response (S–O–R) framework, this study investigates how perceived empathy and interaction modalities influence users’ acceptance intention towards AI-generated health misinformation through underlying psychological mechanisms. A 2 × 2 scenario-based experiment was conducted by manipulating AI empathy (present vs absent) and interaction modality (text vs voice). Survey data collected from Chinese adult users were analysed using analysis of variance and mediation analysis. The results indicate that perceived empathy significantly increases users’ acceptance intention of health misinformation, both directly and indirectly through perceived usefulness and perceived social presence. Interaction modality does not exert a direct effect but influences acceptance intention indirectly via these perceptual factors. The findings reveal a ‘double-edged sword’ effect of empathetic AI in health information contexts: while enhancing perceived value and social presence, empathetic design may also reduce users’ epistemic vigilance and amplify misinformation risks. This study provides theoretical insights and practical implications for the responsible design and governance of empathetic AI systems in health communication.