Deepfakes, the synthetic media generated using artificial intelligence to convincingly depict events that never actually occurred, pose a significant threat to society, yet we still lack efficient and effective countermeasures. While scholars have placed great hope in deepfake priming as a means to combat deepfakes, its effectiveness remains contested and superficial. The nuanced effects of deepfake priming are understudied, particularly regarding how it affects individuals' resilience to deepfakes across media. Therefore, this study examines the interactive effects of deepfake priming and the medium through which deepfakes are delivered on individuals' cognitive and behavioral responses to deepfakes.
We employed an online between-subject experimental design involving 298 US adults. The sample was recruited through a reputable online panel provider, Qualtrics, to ensure a diverse, demographically representative sample. Participants were randomly assigned to one of four experimental conditions in a 2 (deepfake priming vs. no priming) × 2 (news website vs. social media) factorial design. Then, we assessed their belief in the authenticity of the deepfake video and their intention to share it.
We find that priming significantly reduces the intention to share deepfakes but does not enhance the capability to identify them. The medium through which deepfakes are delivered (news website vs social media) does not moderate deepfake priming's influence on either belief in deepfakes or intent to share deepfakes.
This study is among the first to examine the contextual influence of media platforms on deepfake intervention. It enhances the understanding of the deepfake priming effect by exploring the intricate interplay between priming and platforms delivering deepfakes. Practically, it sheds light on developing tailored deepfake interventions and maintaining online information trust and engagement.
