Artificial intelligence and the (in)admissibility of digital evidence secured by the chain of custody:

an analysis in light of the guarantee of evidentiary reliability

Abstract

This work was developed with the aim of analyzing the possibility of presuming digital evidence obtained and discovered by artificial intelligence (AI) in Brazilian criminal proceedings as lawful evidence, thus seeking to understand the role of the chain of custody. Therefore, it sought to present, based on a case, the legal arguments for which judges accepted digital evidence, based on the chain of custody as a mechanism of reliability and its adaptation to the digital environment. Then, it introduced the complex issue of the (in)admissibility of digital evidence obtained by AI, examining the issue of transparency of the methods used. Soon after, it presented the hidden risks of using AI in evidentiary means, such as the lack of interpretability and algorithmic biases, as well as the importance of human supervision and transparency as means of minimizing these risks. In this perspective, through the inductive method, bibliographical research, articles and judgments were carried out to explore the topic. In the end, it was concluded that there is an urgent need for regulations that address the chain of custody of digital evidence, as well as control over the use of AI in criminal proceedings, involving mechanisms that enable the exploration of the potential of AI, and, at the same time, minimize the risks that may affect the rights of the parties.