Are humans able to discriminate between real and deepfake faces? A systematic review and meta-analysis

As deepfakes (i.e., contents generated by artificial intelligence) become increasingly realistic, humans struggle to distinguish them from authentic materials. Deepfake faces have attracted particular attention in this context, given their critical implications for cybersecurity, political misinformation, or identity theft. Considering that human face processing is a highly specialized cognitive mechanism, it raises the question whether this might facilitate the discrimination between this type of deepfake stimuli and real stimuli. The present systematic review and meta-analysis screened the literature for studies involving humans performing deepfake detection tasks on real vs. deepfake static facial stimuli. Including 36 studies (k = 51 experiments), totaling 13197 participants, we tested if humans perform above chance-level in face detection tasks and if accuracy is higher for real (vs. deepfake) stimuli, alongside moderator analyses including a series of methodological variables (i.e., publication year, study quality, deepfake type and response modality). Overall, our results provide evidence that humans perform above chance-level (56.1%), and confirm significantly higher accuracy for real vs. deepfake face stimuli. However, significant heterogeneity observed suggests high variability in the results obtained by individual studies and both effects are shown to be moderated by the assessed variables. Finally, sub-group analyses on experiments aiming to improve deepfake detection indeed reached considerably higher accuracy (62.2%). However, practical relevance of the encountered accuracy should be considered with caution. Our study provides an important contribution to the current knowledge of (deepfake) face processing and illustrates limitations and possibilities, with important applied implications regarding the human ability to identify deepfake (vs. real) faces.