The Distinct Wrong of Deepfakes

Abstract Deepfake technology presents significant ethical challenges. The ability to produce realistic looking and sounding video or audio files of people doing or saying things they did not do or say brings with it unprecedented opportunities for deception. The literature that addresses the ethical implications of deepfakes raises concerns about their potential use for blackmail, intimidation, and sabotage, ideological influencing, and incitement to violence as well as broader implications for trust and accountability. While this literature importantly identifies and signals the potentially far-reaching consequences, less attention is paid to the moral dimensions of deepfake technology and deepfakes themselves. This article will help fill this gap by analysing whether deepfake technology and deepfakes are intrinsically morally wrong, and if so, why. The main argument is that deepfake technology and deepfakes are morally suspect, but not inherently morally wrong. Three factors are central to determining whether a deepfake is morally problematic: (i) whether the deepfaked person(s) would object to the way in which they are represented; (ii) whether the deepfake deceives viewers; and (iii) the intent with which the deepfake was created. The most distinctive aspect that renders deepfakes morally wrong is when they use digital data representing the image and/or voice of persons to portray them in ways in which they would be unwilling to be portrayed. Since our image and voice are closely linked to our identity, protection against the manipulation of hyper-realistic digital representations of our image and voice should be considered a fundamental moral right in the age of deepfakes.