Detect Fakes

Detect Fakes is a research project hosted at Northwestern University by researchers at the Kellogg School of Management, examining how people distinguish truth from fiction in online media as synthetic media becomes more realistic. The program aims to characterize what synthetic media looks like, benchmark how well people can separate synthetic from authentic media across contexts, and develop insights for helping people tell them apart. The work began at the MIT Media Lab in 2019, comparing human, AI, and combined human+AI performance at detecting deepfakes, with a first peer-reviewed paper published in 2022 in the Proceedings of the National Academy of Science (PNAS).

The site offers interactive studies, including a "Same Person or Not?" audio study, a Detect Fake Images or Videos study, and an experiment presenting short social-media texts written by a person or generated by AI, where participants choose which of two accompanying explanations they prefer. Related research topics include human detection of political speech deepfakes across transcripts, audio, and video; characterizing photorealism and artifacts in diffusion model-generated images; and distinguishing AI-generated images from authentic photographs. Participation is voluntary and restricted to those 18 years of age or older, with data collected anonymously for research purposes.