Paper Type
ERF
Abstract
Generative AI intensifies information asymmetry on creative platforms, where audiences prefer human-made artworks but cannot reliably verify origin. We examine how mandatory AI disclosure and metadata-based automated detection shape creators’ opportunistic behavior and audience responses. Leveraging two sequential policy shocks on a digital art platform and employing RDiT, we find that mandatory disclosure initially deters label evasion. However, once metadata-based detection and auto-labeling are introduced, evasion rebounds as creators strategically adapt to these predictable detection rules. Simultaneously, audiences update beliefs: unlabeled works receive greater engagement and sales, suggesting that automated labeling increases the signaling value of non-labels. Our findings reveal a governance paradox in AI-mediated markets: stronger technological enforcement can weaken deterrence by reducing ambiguity and reconfiguring market signals. The study contributes to research on platform governance, deterrence, and disclosure by highlighting socio-technical feedback loops in AI transparency regimes.
Paper Number
1641
Recommended Citation
Du, Tianwen; Xue, Ling; Liu, Xuan; and Song, Peijian, "Cat-and-Mouse in the Gallery: Mandatory AI Disclosure, Detection, and Opportunistic Behavior on an Online Art Platform" (2026). AMCIS 2026 Proceedings. 9.
https://aisel.aisnet.org/amcis2026/ai_systdesign/ai_systdesign/9
Cat-and-Mouse in the Gallery: Mandatory AI Disclosure, Detection, and Opportunistic Behavior on an Online Art Platform
Generative AI intensifies information asymmetry on creative platforms, where audiences prefer human-made artworks but cannot reliably verify origin. We examine how mandatory AI disclosure and metadata-based automated detection shape creators’ opportunistic behavior and audience responses. Leveraging two sequential policy shocks on a digital art platform and employing RDiT, we find that mandatory disclosure initially deters label evasion. However, once metadata-based detection and auto-labeling are introduced, evasion rebounds as creators strategically adapt to these predictable detection rules. Simultaneously, audiences update beliefs: unlabeled works receive greater engagement and sales, suggesting that automated labeling increases the signaling value of non-labels. Our findings reveal a governance paradox in AI-mediated markets: stronger technological enforcement can weaken deterrence by reducing ambiguity and reconfiguring market signals. The study contributes to research on platform governance, deterrence, and disclosure by highlighting socio-technical feedback loops in AI transparency regimes.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
AI SYSTEM