Artificial Intelligence as an Evidentiary Tool: Challenges and Risks in Contemporary Criminal Law

Abstract

This research analyzes artificial intelligence as an evidentiary tool and identifies the risks and challenges in the contemporary criminal justice system. The constant evolution of the world and technology means that the law must keep pace with digital advancements, particularly with evidentiary tools, where artificial intelligence presents opportunities for efficiency, but also substantial risks to procedural guarantees in contemporary criminal law. The methodology employed is qualitative, with an analytical and descriptive scope and a non-experimental design. Bibliographic and doctrinal analysis was used, employing an analytical, inductive, and comparative method, using bibliographic records and inventories as instruments for collecting sources. The results show that the evidentiary reliability of AI depends on data quality, algorithmic transparency, traceability, and the real possibility of rebuttal. Significant risks associated with bias, opacity, biometrics, and deepfakes are evident, with impacts on the presumption of innocence and equality of arms. It is concluded that AI is admissible only as an auxiliary tool for obtaining evidence, under significant human control and reinforced standards of admissibility and evidentiary assessment in order for it to have probative effectiveness.