Deepfakes, Fake Elvises, and Human Interpersonal Relations

Most scholars think that using deepfakes to cause epistemic harm (i.e., to mislead, conceal truth, etc.) may cause us to stop trusting digital testimonies as a whole. This, in turn, will have serious consequences on our interpersonal relations—since most of our communication is conducted in the digital world. I argue that deepfakes created for non-malevolent purposes (e.g., for fun) are epistemically far more dangerous: they undermine digital communication inadvertently and to a much greater extent than malevolent deepfakes. Specifically, they allow malevolent deepfakes to camouflage themselves as non-malevolent (i.e., blend with the environment) thus creating a context of epistemic uncertainty. However, we may have reasons to believe that rational self-interested agents will agree to substantially limit or ban the use of deepfakes. Rather than educating people on the nature of deepfakes or misinformation, we should educate them on the potential harm they could cause to others and themselves by creating and disseminating deepfakes.