Deepfakes: Definition of the Concept and Criteria for Distinguishing Between Harmful and Harmless Deepfakes
Deepfakes: Definition of the Concept and Criteria for Distinguishing Between Harmful and Harmless Deepfakes
A. Hachkevych
TLDR
The relationship between the dissemination of harmful deepfakes and relevant legal categories such as revenge porn, defamation, right to privacy, unfair competition, and disinformation known in contemporary legal systems are discussed.
Abstract
Abstract. This article addresses the issue of combating deepfakes, which has recently gained significant relevance. With the emergence of publicly available artificial intelligence tools capable of generating highly convincing images, video clips and other types of content, as well as a favorable digital landscape for their dissemination, deepfake technology has become increasingly prevalent. Given the risks of deepfakes, reasonable expectations are placed on the law designed to protect our fundamental values, which are often
jeopardized by deepfakes. The spread of harmful deepfakes poses risks to the individuals depicted, causes damage and destroys the reputation of affected organizations, and can be dangerous for society by serving as an effective means of disinformation and manipulation of public opinion. Therefore, the author examines deepfakes as a challenge, exploring the concept in depth and highlighting some of the contentious issues. This article outlines four criteria to differentiate harmful deepfakes from harmless ones: a) consent of the individual–whether a person featured in a deepfake has agreed to its creation and dissemination, b) absence of criminal acts–whether a deepfake involves any illegal activities, c) indication of artificial intelligence usage– whether a deepfake clearly demonstrates characteristics of being created by artificial intelligence, d) understanding of their intents considering serving a social good. The author also attempts to provide his own definition of the concept of a deepfake, and to outline its components: technological and intellectual. Alongside traditional images and videos, he considers that audio recordings should be regarded as a type of deepfakes. Furthermore, this article discusses the relationship between the dissemination of harmful deepfakes and relevant legal categories such as revenge porn, defamation, right to privacy, unfair competition, and disinformation known in contemporary legal systems.