Preventive Criminal Policy and the Challenges of Attributing Criminal Liability in the Production and Dissemination of Sexual Deepfakes

Abstract

The emergence of novel technologies, particularly generative artificial intelligence and the phenomenon of “deepfakes,” has created fundamental and unprecedented challenges for criminal justice systems. Among these challenges, the production and dissemination of fabricated pornographic content have increasingly targeted individuals’ reputation, privacy, human dignity, and psychological security on a large scale. In confronting this phenomenon, the traditional architecture of criminal law and existing legislation, including the Computer Crimes Act of 2009, have become structurally inadequate due to their predominantly ex post approach and the absence of differentiated criminalization tailored to the specific characteristics of deepfake-related conduct. The present study seeks to identify legislative gaps and propose an appropriate regulatory and preventive model for addressing this phenomenon. Using a descriptive-analytical method and drawing upon library-based sources, the study examines the legal and criminological dimensions of sexual deepfakes. The findings indicate that, within the decentralized and networked environment of cyberspace, an exclusive judicial focus on punishing end users—whether producers or disseminators—is inconsistent with the logic of criminal justice economy, particularly given the extensive and diffuse nature of potential victimization. Overcoming this crisis requires a paradigm shift from a “repressive criminal policy” toward a “participatory and preventive regulatory model.” Accordingly, criminal liability should be fairly and purposefully distributed across three levels: users, dissemination platforms, and technology developers. Abandoning a passive notice-based approach and imposing “enhanced corporate liability” on hosting platforms for proactive and algorithmic content filtering, together with establishing a regime of “liability arising from the design of high-risk digital products” for developers—including obligations to embed covert watermarking mechanisms and technological safeguards—constitute key strategies of situational and technical prevention. Ultimately, this article argues that effectively controlling deepfake-related offending requires the enactment of comprehensive artificial intelligence legislation, the adoption of active legal diplomacy at the international level, and the reciprocal use of defensive artificial intelligence in the scientific detection and investigation of crimes.