Seeing through the fake: how users detect and interpret deepfakes

Deepfakes pose significant challenges to information credibility and user discernment. While public and scholarly interest in deepfakes has grown, little is known about the cognitive processes that shape individuals’ judgments of their credibility. This study investigates how users assess deepfakes by examining the role of heuristic cues and contextual features in influencing evaluative judgments. Drawing on dual-process theories of cognition, the study develops a model that explores how heuristic processing (e.g., bandwagon cues, source attribution) and perceptions of verisimilitude affect users’ systematic decision-making and behavioral intentions, particularly regarding the sharing of deepfake content. Findings reveal that users who perceive deepfakes as highly realistic are more likely to evaluate them systematically and consider sharing them. Reliance on cognitive shortcuts often overrides analytical reasoning. Notably, users’ ability to discern the authenticity of deepfakes–diagnosticity–improves when they are exposed to relevant cues and transparent explanatory information. This study enhances our understanding of how individuals assess the credibility of deepfakes and lays the groundwork for developing interventions to counter the spread of misinformation.