Beyond the Transparency Paradox: Contextual AI Disclosure and User Trust on Chinese Digital Platforms

Abstract

This study explores how varying degrees of disclosure of artificial intelligence information affect users' trust and behavioral intentions on Chinese digital platforms. Based on recent controlled experiments, eye-tracking research, and statistics, this paper finds that the disclosure of artificial intelligence information can have a double-edged effect, depending on the specific situation. In utilitarian situations such as advertising and product search, information disclosure tends to improve user trust and perceived credibility. In social and hedonic situations, information disclosure tends to reduce user engagement and weaken the connection between users and content. Mechanism analysis shows that the driving factors of this pattern include: first, users' attribution bias toward errors labeled as AI-generated. Second, the increase in effectiveness and discomfort after information disclosure, and finally, the non-linear relationship between the level of information disclosure detail and trust. These mechanisms vary depending on the platform type, product type, and the user's artificial intelligence literacy. Based on these findings, this study proposes a three-part governance framework: a tiered information disclosure system, traceable records that meet the current regulatory requirements, and a human-in-the-loop design for high-sensitivity interactions. The framework challenges a common assumption that more information is better in artificial intelligence information disclosure. On the contrary, it adopts a different approach - a situation-oriented approach. The goal is to achieve a balance among transparency, user trust, and user participation on digital platforms.