Deepfakes as narratives: Psychological processes explaining their reception

Abstract

Hyper-realistic AI-generated synthetic media, known as deepfakes, manipulate appearances and actions, often to depict fabricated events. As technological advancements make them more and more realistic, their potential for both harmful deception and creative innovation becomes stronger than ever. This study explores how individuals interact with deepfakes of different genres when they are aware of their artificial nature. It proposes a model that connects these reactions to the psychological processes that are typically involved in responses to fictional narratives. A total of 1,031 US participants (MAge = 43.0, SD = 14.0, RangeAge = 18–93) watched one of ten randomly assigned deepfakes (5 genres x 2 different videos) and then completed a series of questionnaires assessing familiarity with the protagonist, narrative transportation, perceived realism, message evaluation, enjoyment, and sharing intentions. The results confirm that, regardless of genre, narrative transportation prompts positive evaluations of perceptual quality and narrative consistency, which in turn influences enjoyment. Greater enjoyment, transportation and familiarity with the protagonist are predictors of heightened sharing intentions. These findings address gaps in our understanding of how people respond to deepfakes, with implications for entertainment and narrative theory.