How deepfake plausibility, media literacy, and personal attitudes shape detection, liking, and intentions to share political deepfakes on social media

The increasing realism and accessibility of political deepfakes pose serious risks to democratic discourse by blurring the line between authentic and manipulated media. The present study examined how deepfake plausibility, media literacy, and attitudes toward the person depicted in the video shape individuals’ ability to detect political deepfakes, their emotional responses (liking), and behavioral intentions (social media sharing intentions). A total of 1124 participants from the United Kingdom, Slovenia, and Italy viewed manipulated videos about climate change and immigration that varied in plausibility (technical realism and content alignment with the speaker’s publicly known stance).

The results showed that high-plausibility deepfakes were less likely to be detected and received more positive evaluations than low-plausibility ones. In contrast, plausibility did not significantly influence sharing intention. Media literacy and attitudes toward the person in the video emerged as strong predictors across outcomes; higher media literacy was associated with improved detection and with reduced liking and sharing intention, while more positive attitudes were associated with reduced detection and with increased liking and sharing intention.

Mediation analyses demonstrated that liking partially mediated the link between detection and sharing intention. Furthermore, moderated mediation models revealed that this indirect effect was stronger among individuals with high media literacy (for climate change videos) and favorable attitudes toward the video’s subject (for both topics). Overall, the study highlights the need for interventions that address not only detection skills but also emotional and motivational susceptibility to persuasive synthetic media.

Enhancing media literacy and implementing platform-level friction mechanisms may help curb the spread of political deepfakes online. The authors would like to thank other members of the project consortium, particularly the project leader, Dr. Federica Russo, for fruitful discussions and coordination efforts.

We would also like to thank psychology students who helped collect data in Slovenia (listed alphabetically): Evgenija Avtarovska, Špela Brezočnik, Pia Gričar, Ria Grof, Katarina Hlede Jelen, Anisa Kohne, Eva Krošel, Kiara Marinič, Maša Podvršnik, Ratka Tancheva, and Simon Žnider. The authors declare that financial support was received for the research, authorship, and/or publication of this article. This research was funded by the project ‘SOLARIS: Strengthening democratic engagement through value-based generative adversarial networks’ that has received funding from the European Union’s Horizon Europe research and innovation program (GA No. 101094665).

The content of this paper does not reflect the official opinions of the funders or any other institution. The responsibility for the information and views expressed herein lies entirely with the authors. Department of Psychology, University of Maribor, Maribor, Slovenia Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor, Slovenia Research Unit on Theory of Mind, Department of Psychology, Catholic University of the Sacred Heart, Milan, Italy Sapienza University of Rome, Rome, Italy The authors declare no competing interests.