When Trust Does Not Translate into Engagement: A Decoupling Perspective on AI Disclosure in Short-Form Video Platforms

The rapid growth of AI-generated content on social media raises important questions about how transparency mechanisms, such as AI disclosure, influence user behavior.Prior research reports mixed findings, suggesting that disclosure may reduce trust but has inconsistent effects on engagement.We propose that these inconsistencies can be explained by a decoupling between evaluative judgments and behavioral responses, whereby trust and engagement do not necessarily move together.Drawing on attribution theory, we examine how disclosure source (platform vs. creator) shapes user responses through trust in content and creator.Using a controlled experiment on short-form videos, pilot results show that AI disclosure significantly reduces trust, particularly when generated by the platform.However, disclosure does not directly reduce engagement, providing initial evidence of decoupling.This study contributes by introducing a decoupling perspective to AI disclosure research and highlighting the importance of disclosure source in shaping user behavior.