Abstract
Deepfakes pose a particularly sophisticated threat due to their vivid realism and high credibility. Yet, empirical research about how Deepfakes affect business and brand is infertile. While warning labels are a common mitigation measure on social media platforms, current practices predominantly rely on plain-text formats, often mismatched with the multimodal nature of Deepfakes. This modality mismatch may undermine the intervention effect. Additionally, the argument framing within a warning label significantly influences its perceived credibility and ability to counteract Deepfakes. This study investigates how modality (text, image, video), argument framing (forensic, users’ social network-based, content-based), and their interaction affect the warning label's efficacy, in terms of attitude and behavior toward the target brand. We anticipate our findings will offer actionable insights for social media platforms to implement more effective measures against Deepfake disinformation.
Recommended Citation
Chen, Wei-Chang and Mirhoseini, Mahdi, "The Influence of Modality and Argument Framing of The Warning Label Design on Counteracting Deepfakes" (2025). SIGHCI 2025 Proceedings. 4.
https://aisel.aisnet.org/sighci2025/4