A inteligência artificial no direito penal e os novos desafios aos direitos fundamentais

Abstract

This study critically analyzed the impact of artificial intelligence on criminal law and procedure, paying special attention to predictive justice and the use of digital evidence, as well as the resulting challenges to fundamental rights. The research began with an examination of Brazilian laws that indirectly address the topic: the Consumer Protection Code (Law 8.078/90), the Marco Civil da Internet (Law 12.965/2014), and the General Data Protection Law (Law 13.709/2018). Due to the significance of this issue, it was determined that a separate law addressing AI is necessary, referencing Bill No. 2,338/2023. After analyzing the regulatory landscape, the paper discussed predictive justice, its advantages, and its disadvantages. It highlighted that the opaque and complex nature of algorithms can compromise the fulfillment of fundamental rights in decision-making processes. The paper also addressed criminal liability in relation to autonomous vehicles and questioned the criteria used to program systems that must make a "life vs. life" choice. The possibility of applying the exclusion of illicitness in relation to the state of necessity was analyzed and found to be inapplicable. A comparison was also made with the Portuguese and German Penal Codes, which, unlike the Brazilian Penal Code, include the concept of "conflict of duties." This study revealed that, in the event of the deliberate programming of tragic choices, depending on the specific case, the programmer's conduct may fall under the culpability exclusion of the inexigibility of different conduct. Regarding digital evidence, the privacy and broad defense risks that can arise from technologies such as malware and the admissibility of such evidence in the Brazilian legal system was discussed. In short, the study has shown that, despite the benefits of using AI in criminal law, caution must be exercised to ensure that fundamental rights are not violated