Not all harms are equal: an investigation of third-person perception of deepfakes harms

Third-person perception (TPP) has been observed across various media contexts. However, little is known about how TPP manifests across different deepfake harm types and age groups. This study aims to investigate this perceptual gap in deepfake harms, as well as the role of prior encounters with dubious content as a moderator. We conducted a survey with three age groups: students from Generation Z, working adults, and seniors, to examine their perceived susceptibility to five types of deepfake harms: physical, emotional, relational, financial and societal harms. Further, we explored the moderating role of prior encounters with dubious content in shaping the magnitude of TPP. Across all age groups and harm types, participants perceived themselves as less susceptible to deepfake harms compared to others. However, the magnitude of this bias differs. Seniors, for example, showed the largest self–other gap, particularly for emotional harms, while societal harms were perceived more equally across groups. Prior encounters with dubious content also played a role. More exposure tended to reduce TPP for physical and financial harms, but interestingly, working adults displayed the opposite trend, with exposure increasing their TPP. These results extend the body of knowledge in TPP research by demonstrating how age, harm type, and prior encounters with dubious content shape perceptions of susceptibility. Practically, the findings highlight the need for tailored media literacy and intervention strategies, especially for seniors who may underestimate their susceptibility.