Active Inoculation Against AI-Generated Misinformation

As worries mount over AI-generated misinformation, questions remain about its influence on people’s thinking and how this influence can be mitigated. The present study investigated the efficacy of an active, versus passive, source-focused inoculation to reduce trust in AI-generated information and its effects on relevant beliefs. Participants (N = 400) read either a misleading AI-generated article about the dangers of nanoparticles in sunscreen, or a neutral control article. The misleading article was either preceded by a passive inoculation explaining how AI-generated information can be misleading, an active inoculation that additionally included an educational activity, or no intervention. The misleading AI article significantly influenced participants’ beliefs about nanoparticles in sunscreen. Both passive and active inoculation reduced general trust in AI-generated information, as well as the impact of the specific misinformation on beliefs, with active inoculation demonstrating larger effects; however, neither intervention eliminated the influence of misinformation on beliefs. These findings suggest that source-focused inoculation is more effective at reducing the impacts of AI-generated misinformation when it incorporates an active component. However, the continued influence of the misinformation even in the active-inoculation condition calls for alternative or combination interventions.