Privacy concerns about third-party images hosted by deepfake technologies

The emergence of deepfakes, also referred to as Artificial Intelligence Generated Contexts (AIGC), has sparked significant scholarly and public interest. Deepfakes have both captivated and raised concerns among observers due to their profound implications for media productivity and manipulation, which in turn challenges traditional concepts of reality. Additionally, the ethical and legal implications surrounding the creation and dissemination of deepfakes have become subjects of intense scrutiny.

Leveraging artificial intelligence (AI) and computer graphics, deepfakes have rapidly evolved, enabling the creation of realistic videos, audio, and images portraying events and scenarios that are entirely fictitious. A creator of a deepfake video uses an image of a third party to create content that may depict the person’s image and voice doing various acts. The creation of a deepfake video may use a person’s personal data such as image, voice, and other personal data without their permission. This personal data may fall within the scope of privacy law; however, when an individual creates a deepfake depicting third-party images and voice data, it potentially falls outside the scope of protection of privacy law.

Aside from privacy issues, there are concerns that regulation of deepfakes might clash with freedom of expression. In some circumstances where there are gaps in privacy protection, tort law may fill in some gaps where there has been an unlawful use of one’s information for deepfakes. Although in the absence of one’s clear consent to the use of images, no exchange takes place, courts might not stand behind all litigations.

Recently, some big data companies have scraped all public images and data to train their artificial intelligence large language models, which raises questions of consent for the use of those images. However, the further use of images to create deepfake content can amount to an involuntary exchange. In this respect, some activities organized by voluntary exchanges can lead to wealth maximization, meaning that individuals have access to required materials to increase their productivity.

Despite this, some people are willing to bypass the market through coercive actions. In such cases, a transaction is not concluded, nor can compensation be paid to the influenced parties. With a distortive impact on the efficient allocation of resources, society is prompted to criminalize these actions. Hence, social institutions should deal with two broad types of behaviors, legal and criminal.

Although society aims to crack down on criminal actions, transgressors are still looking for new ways to reap the benefits without obeying the rules of the market. In this vein, deepfake technology instrumentally assists one with exploiting and jeopardizing others’ information and reputation, respectively, which in turn disrupts the legitimate market for deepfakes.

This paper aims to examine the legal status quo and the recent legal movements in Australia, the European Union, and the United States regarding privacy concerns arising from the prevalence of deepfake technologies. However, litigations might not effectively deter creators or distributors of illegal deepfake content. Thus, an ex-ante prohibitive law can be regarded as a substantial constraint on deepfake technologies with a sound safeguard for privacy. In this vein, it is assumed that the prohibition of destructive uses and the promotion of instructive applications of deepfakes are possible through legal instruments.