O uso de Inteligência Artificial na Produção de provas no Processo Penal Brasileiro: desafios e perspectivas para a preservação dos Princípios Processuais Fundamentais

Abstract

The dissertation addressed the use of Artificial Intelligence (AI) in the context of criminal proceedings, aiming to enhance efficiency and security in the production of evidence while respecting fundamental procedural principles. The research adopted a deductive approach and relied on documentary and bibliographic research, structured into three chapters with several sections. The first chapter establishes the theoretical foundations of Artificial Intelligence and its application in the legal context. Artificial Intelligence is defined as the ability of machines to mimic human intelligence, employing methods such as supervised, unsupervised, reinforcement learning, and natural language processing. The research underscores the importance of ensuring that the use of this technology in criminal proceedings complies with legal and constitutional principles. The second chapter explores the primary uses of Artificial Intelligence in criminal matters. This includes digital data analysis, emphasizing the need to balance the use of big data and AI with the General Data Protection Law (LGPD) to ensure compliance with principles such as purpose and transparency. Facial recognition is addressed, highlighting privacy concerns and the risks of identification errors. Prediction of judicial decisions is discussed, with an emphasis on the restrictions imposed by the National Council of Justice (CNJ) and the importance of transparency and human review. Behavioral analysis and DNA analysis are presented as evidence production techniques in criminal proceedings. The third chapter examines the implications of AI use in criminal proceedings, emphasizing the importance of legal regulation to oversee the changes brought about by digitization and the need to balance efficiency with the protection of individual rights. The principle of adversarial proceedings is addressed, emphasizing the need to ensure equal access to justice in a context where Artificial Intelligence can analyze large volumes of data. Privacy is discussed, highlighting the importance of LGPD and similar regulations to protect personal data. The research also underscores the relevance of digital chain of custody to ensure the authenticity and integrity of evidence obtained through Artificial Intelligence. In addition to the risks related to privacy and transparency, the use of Artificial Intelligence in criminal proceedings may also be subject to algorithmic bias. Algorithms can exhibit bias if trained on biased data, potentially leading to unjust judicial decisions. The dissertation's conclusion emphasizes that the use of AI in criminal proceedings is a real possibility, although its scope in Brazil is limited by the CNJ, which prohibits the use of predictive judicial decisions in criminal matters. This is considered reasonable since there is no clear regulation with mandatory compliance principles to ensure transparency, privacy, and limits on its use.