Exploring the diffusion mechanism of generative AI disinformation in online platforms: an explanatory model

Purpose There is currently no research analyzing the mechanisms of the spread of AI-generated disinformation to curb the spread of disinformation. This study investigates how generative AI influences disinformation dissemination within social networks, with a particular focus on the diffusion mechanisms. Design/methodology/approach The study analyzes interaction data between AI and users on two well-known social platforms: Zhihu and Facebook, employing mixed-methods to examine the characteristics and influencing factors of different types of AI disinformation. Findings The results indicate that the diffusion of AI-generated disinformation is influenced by network structure and node filtering. Additionally, the research introduces the “Bee” diffusion model, which in detail presents the pathways and dynamics of AI disinformation spread within social networks, providing new perspectives and tools for understanding and controlling AI disinformation. The “Bee” diffusion model illustrates how AI-generated content alters social network structures and influences user interactions, providing theoretical guidance on regulating and optimizing information dissemination models. Originality/value Due to this model’s predictive nature, it can help researchers predict the path of AI disinformation. By analyzing the behavioral patterns of accounts, especially the data of their social media interactions, this model is able to more accurately predict the spread trend of AI disinformation. On the one hand, identifying AI accounts in the communication chain based on AI signs can help find the source of disinformation. In addition, by taking a rough portrait of users (e.g., gender, work, and social network, etc.), this model is able to identify key nodes that may be affected by the disinformation and predict its path. Peer review The peer review history for this article is available at: https://publons.com/publon/10.1108/OIR-09-2024-0595