Seeing and hearing is no longer believing.
Generative AI can fabricate realistic visual imagery, clone a person's voice from seconds of audio, and produce convincing "enhancements" of low-quality evidence — all with free or low-cost tools. These capabilities have already surfaced in misidentifications, in election-interference robocalls, and in evidence introduced in criminal trials.
Courts, legislatures, journalists, and the public are confronting these challenges, often without scientific grounding. Project PROVE exists to build the evidence base they need. We aim to conduct rigorous, independent research and evaluation at the intersection of perceptual science, computational analysis, forensics, and the law.
