Effects of inoculation strategies on perceived accuracy and sharing intentions across deepfake modalities

Abstract

Deepfakes are audiovisual manipulations that are increasingly pervasive on the internet and commonly weaponised for malicious intents that sow distrust. Thus, the need to safeguard susceptible individuals is ever more paramount. Study 1 aims to identify characteristics that affect an individual’s perceived accuracy and sharing intentions of deepfake videos. Social media fatigue and dogmatism were found to have positive associations with such outcomes, while feelings of control and cognitive ability had negative associations. Study 2, our follow-up study, investigates the use of inoculation to safeguard against deepfake images and videos. This involves harms-based (i.e., general information about deepfakes that do not aid in identifying deepfakes) and medium-specific identification-based strategies (i.e., modality-specific information that aids in identifying deepfakes). Results revealed that the presence of inoculation reduced perceived accuracy when compared to controls, but only for deepfake images. Inoculation did not reduce sharing intentions for both deepfake images and videos. The mediating role of perceived accuracy between the presence of inoculation and deepfake concerns was also explored. Our findings not only contribute to the scarce research on deepfakes and inoculation, but also inform the development of more effective inoculation measures to reduce susceptibility to disinformation.