Investigating Disinformation with Virtual Influencers

This study investigates how users perceive and engage with disinformation when it is disseminated by virtual influencers (VIs), a growing category of AI-generated personas active on social media. While previous research has primarily focused on political figures or human influencers, this study examines responses to a non-political, human-like VI and includes digital literacy as a moderating factor. An online experiment was conducted using a between-subjects design (N = 220), in which participants were randomly assigned to view an Instagram-style post from the VI "Lil Miquela," featuring either a neutral caption or a fabricated disinformation caption generated using AI. Participants then rated the perceived accuracy of the post and their intention to share it. Measures of digital literacy were also collected. We find that individuals exposed to a disinformation caption from a VI are more likely to perceive the claim as inaccurate compared to those shown with a neutral caption. Individuals are also less likely to share the post when they judge the claim to be less accurate. Moreover, we find that digital literacy plays a moderating role such that individuals with higher digital literacy are less influenced by the disinformation caption, showing lower perceived accuracy in response to misleading content.