The goal of this work, developed by researchers in the Affective Computing and Viral Communications research groups, is to "prebunk" deepfakes—videos manipulated by neural networks—by.