The Admissibility of Evidence Obtained using Artificial Intelligence Techniques in Customs Crimes and Cases of Invalidity

Abstract

This research aims to examine the admissibility of evidence obtained through artificial intelligence technologies in customs-related crimes, with a particular focus on the legal and procedural safeguards that ensure the validity of such evidence and the circumstances that may lead to its nullification. The study seeks to establish a normative framework that balances technological innovation with the guarantees of criminal justice, thereby enhancing the judiciary’s ability to address complex customs offenses that increasingly rely on digital and intelligent methods.

The significance of this research lies in its exploration of a contemporary and interdisciplinary topic at the intersection of criminal law and technology. Customs crimes have become more sophisticated with the rise of digital smuggling and logistical manipulation, while reliance on artificial intelligence in detecting such crimes raises legal challenges concerning legality, interpretability, and fairness. Thus, the study provides an original contribution to modern legal scholarship and offers practical solutions to the challenges facing customs adjudication. The research adopts a descriptive, analytical, and comparative methodology. It analyzes national and international legal texts related to digital and customs evidence, examines practical applications of artificial intelligence in customs investigations, and compares judicial experiences across jurisdictions to identify strengths and weaknesses. This methodological approach enables the extraction of standards of admissibility and grounds for nullification through a critical analysis of legal and technical literature. The findings reveal that evidence derived from artificial intelligence in customs crimes enjoys conditional admissibility, provided that its collection is lawful, its digital chain of custody is properly documented, and its reasoning is subject to judicial interpretation. Such evidence cannot be considered conclusive on its own but rather serves as a supporting technical indication that must be corroborated by traditional forms of proof. Grounds for nullification include the absence of legal authorization, defects in the chain of custody, reliance on opaque algorithmic models that hinder the right of defence, algorithmic bias, and violations of privacy.