In 2026, the Semarang District Court adjudicated a case illustrating how artificial intelligence can be misused for criminal purposes. The defendant was found guilty of producing and disseminating more than 300 items of AI-manipulated pornographic content commonly referred to as deepfakes. This technology manipulates a person’s image or video to create highly realistic yet entirely fabricated content, potentially causing victims prolonged psychological trauma and persistent social stigma. The legal complexity intensifies when such cases reach the courtroom because digital evidence is inherently susceptible to manipulation, duplication, and deletion. This article analyzes the judicial reasoning employed in assessing digital evidence and examines its evidentiary value in Decision No. 12/Pid.Sus/2026/PN Smg. The study adopts a normative juridical method using statutory, case, and conceptual approaches, with the court decision serving as the primary source of legal material and without relying on interviews or direct observation. The findings indicate that the judges adequately correlated digital evidence with witness testimony, physical evidence, and the defendant’s confession through an evidentiary construction encompassing juridical, philosophical, and sociological considerations. When assessed against Debra L. Shinder’s four criteria for the admissibility of electronic evidence, the digital evidence in this case satisfied the requirements of relevance and materiality, while its reliability and credibility were reinforced through consistency with other forms of evidence. However, technical authentication could not be fully assessed because the decision did not elaborate on metadata examination, hash values, digital forensic procedures, or chain-of-custody mechanisms, and no digital forensic expert was presented at trial. These findings underscore the need for regulatory harmonization and more specific legal provisions criminalizing deepfake pornography as a distinct criminal offense.