Abstract

The rise of AI-Generated Content (AIGC) raises concerns about deceptive content and trust erosion, necessitating effective disclosure policies. Despite the emerging practices, little is known about how AIGC disclosures affect user behaviour and platform dynamics. This study investigates the impact of AIGC disclosure on video consumption and user engagement using data from a prominent online video platform. We employ a Difference-in-Differences (DID) methodology to compare categories with high and low AIGC prevalence. Preliminary findings indicate a general decline in plays and likes for AIGC videos post-policy implementation. However, labelled AIGC videos, no matter disclosed by their creator or the platform, receive more engagements than non-labelled counterparts, suggesting that explicit disclosure alleviates user distrust. These insights are crucial for developing effective governance strategies for AIGC and guiding content creators' disclosure practices. Future research will address endogeneity issues and employ advanced causal machine learning techniques to enhance causal inferences.

Download

Share

COinS