The article aims to analyse the applicability of the categories of the general part of Brazilian criminal law to crimes committed by, through and/or to the detriment of Artificial Intelligence (AI) systems. To achieve this purpose, the research, still in progress, was based on exploratory, bibliographic, and indirect documentary research, based on the analysis of interdisciplinary literature within the criminal and computer sciences, considered relevant, as well as within the legal system, through legislation and jurisprudence available. In the development, the current definition of AI system in Brazilian legislation, jurisprudence and doctrine was analysed, verifying also issues related to machine learning and the possibility of existence of legal personality of AI systems. As a partial result, it was obtained that, besides the fact that the Brazilian criminal policy regarding AI-related crimes is embryonic, the latent gap in the legal system regarding the issue was observed, also reflecting the absence of concrete cases that stimulate the application, creation and/or alteration of rules on AI. In this sense, it was suggested to stimulate the legal debate, for the formulation of new scientific research and doctrinal studies on the subject.