When an Influencer is Disclosed as AI: Anthropomorphism and identity Disclosure Shape Engagement Intentions

As AI-generated influencers increasingly resemble human creators, an important question for social media platforms is whether disclosing an influencer’s AI-identity promotes transparency at the cost of user engagement. This research examines whether the effect of AI-identity disclosure on platform-native engagement intentions depends on an influencer’s level of anthropomorphism and through which psychological processes this effect unfolds. Across two mixed-design experiments, we manipulated anthropomorphism (high vs. low) between subjects and measured engagement intentions before and after disclosure within subjects. In Experiment 1, highly anthropomorphic AI-generated influencers initially elicited stronger engagement intentions, but this advantage diminished after disclosure due to a large decrease in perceived humanness. In Experiment 2, a human influencer was labeled as AI to isolate the identity-cue component of disclosure. The same interaction pattern emerged, but the decline in engagement was explained by a stronger reduction in perceived attractiveness. Supplementary analyses further indicated that mediation differences were better explained by participants’ subjective AI attribution, suggesting that disclosure operates through different updating processes depending on how the target is initially construed. These findings extend understanding of how AI-identity disclosure shapes users’ responses to humanlike influencers by identifying both its boundary condition and its underlying psychological pathways.