Immunizing the Public Against AI-Generated Disinformation: Testing the Effects of Inoculation Mode and Issue Attitude on Inoculation Likelihood of Political Deepfakes.

This article examines the effectiveness of inoculation theory-based interventions—specifically passive and active inoculation modes—in increasing resistance to political deepfakes, which are AI-generated synthetic videos that distort politicians' statements and embed disinformation. Through a large-scale online experiment with a demographically representative U.S. sample (N=1,013), the study finds that both passive (text-based) and active (interactive, quiz-based) inoculation increase deepfake awareness, reduce perceived credibility of deepfake videos, lower agreement with embedded disinformation, and enhance intentions to debunk and seek information about deepfakes compared to no inoculation. However, active inoculation uniquely increased information-seeking behavior more than passive inoculation. The study also reveals that exposure to counter-attitudinal deepfakes—videos opposing participants' preexisting abortion attitudes—led to greater agreement with disinformation, illustrating the moderating role of motivated reasoning and issue-attitude congruency on inoculation effectiveness. These findings suggest that while inoculation strategies can bolster public resilience against political deepfakes, their impact varies depending on individuals' prior beliefs and the alignment of the deepfake content with those beliefs.