A Comparative Legal Study on the Criminal Justice Applications of Deepfake Technology: The Feasibility of Introducing Victim Video Testimony in South Korean Sentencing Procedures PARK, SO HEE Advisor : Prof. Kim, Jong Goo, Ph.D. Department of Law, Gra...
A Comparative Legal Study on the Criminal Justice Applications of Deepfake Technology: The Feasibility of Introducing Victim Video Testimony in South Korean Sentencing Procedures PARK, SO HEE Advisor : Prof. Kim, Jong Goo, Ph.D. Department of Law, Graduate School of Chosun University This study examines the admissibility and legal limitations of using deepfake technology in victim impact statements during the sentencing phase of criminal trials. With the rapid advancement of artificial intelligence, particularly generative AI and deepfake technology, new possibilities have emerged across various fields, including the criminal justice system. In particular, when the victim is deceased, the current victim impact statement system faces inherent limitations in conveying the victim’s own voice and experience. In this context, the use of deepfake technology has been suggested as a potential alternative. This study first analyzes the concept and characteristics of artificial intelligence and deepfake technology, followed by an examination of their applicability within the criminal justice system. It then explores the legal structure and function of victim impact statements, focusing on the gap in representation when the victim is deceased. Furthermore, the study reviews the Christopher Pelkey case in the United States as a representative example of the use of AI-generated victim statements in court. The findings indicate that deepfake-based victim statements may enhance the representation of the victim’s perspective and contribute to victim-centered justice. However, such statements raise significant legal concerns, including the lack of evidentiary admissibility, difficulties in verifying authenticity, potential infringement of the defendant’s right to confrontation, and the risk of emotional bias affecting judicial decision-making. Accordingly, this study concludes that the full-scale adoption of deepfake-based victim statements is not appropriate. Instead, their use may be considered only in a limited manner as supplementary materials during sentencing, under strict procedural safeguards. The study further suggests the need for clear legal classification, transparency in production, protection of the defendant’s rights, control of emotional influence, and safeguards for posthumous personality rights. Ultimately, this study contributes to the ongoing discussion on how to balance technological innovation with the fundamental principles of fairness and objectivity in criminal justice.