On July 7, 2026, researchers from Lancaster University published a study revealing that images of faces created by artificial intelligence (AI) are perceived as more trustworthy than real human faces. This groundbreaking research highlights the potential dangers of online fraud and misinformation.
Study Overview and Methodology
The study, led by Alexis McGuire and involving researchers from Stanford University and the University of California, Berkeley, is the first to investigate the trustworthiness of AI-generated faces using advanced diffusion technology. The team conducted experiments with 169 participants who evaluated a collection of 96 diverse faces, assessing whether each was real or AI-generated.
Participants displayed an average accuracy of only 58.4% in identifying real versus AI faces, which is only slightly better than random guessing. Surprisingly, faces generated by the newer AI diffusion model were rated as less realistic than those produced by an earlier model known as GAN. However, in terms of trustworthiness, diffusion-generated faces were rated the highest.
Trustworthiness Ratings of AI and Real Faces
In a follow-up experiment, participants rated the trustworthiness of the same 96 faces on a scale from 1 (very untrustworthy) to 7 (very trustworthy). The findings revealed:
- Real faces: average trust rating of 4.03
- GAN faces: average trust rating of 4.36
- Diffusion model faces: average trust rating of 4.70
These results indicate that both AI-generated faces were deemed more trustworthy than real faces, raising concerns about societal implications as AI technology continues to advance.
Implications of AI-Generated Images
McGuire emphasized the paradox in the findings, stating, "This finding presents a paradox and thus highlights the possibility that realism and trustworthiness judgments are driven by two different psychological mechanisms." The researchers warned that the increasing sophistication and accessibility of AI-generated images could undermine public trust, contributing to challenges such as identity fraud, catfishing, and political disinformation.
As AI faces become more prevalent, it is crucial for society to recognize the potential risks associated with their use and to develop strategies for mitigating these harms. The research is published in the Journal of Vision and encourages participation in an anonymous survey aimed at exploring individual differences in detecting real versus AI-generated faces.
🤖 This article was rewritten by Feed and Figures' editorial AI from a report originally published by Phys.org. Facts and quotes are preserved from the original; the rewrite focuses on clarity and structure. For the unedited original, see the source link below.
