Using artificial intelligence for the preventive detection of forensic risks in entrepreneurship: an interdisciplinary approach

This article presents a comprehensive study of innovative methods of criminal law counteraction to crime in the sphere of entrepreneurship under the conditions of the digitalization of the legal system. Particular attention is paid to the integration of modern information technologies, especially artificial intelligence (AI) tools, into the activities of law enforcement agencies for the purposes of preventing, detecting, and investigating economic offenses. The paper emphasizes the application of the concept of "predictive policing," which involves the use of machine learning algorithms to generate criminological risk models and identify potentially harmful entrepreneurial activity before a crime is committed. The study also addresses the issue of legal admissibility of AI-generated analytical results in criminal proceedings. It highlights the potential risks associated with erroneous conclusions of neural networks (so-called "AI hallucinations") and the difficulties of legal assessment of outcomes generated by opaque algorithms (the "black box" effect). The necessity of developing normative legal mechanisms is substantiated, which would establish criteria for the admissibility, reliability, and verification of analytical information generated with the involvement of artificial intelligence. The article concludes that the system of "predictive policing" – i.e., the implementation of technologies for preventive surveillance aimed at forecasting future crimes and individuals potentially capable of committing them, as well as forming profiles of such persons and potential victims – does not comply with the principles of criminal procedural law of Ukraine. In particular, this model contradicts the prohibition of objective imputation – that is, the inadmissibility of holding a person criminally liable for actions or consequences not resulting from their conscious intent or will. Proactive criminal procedural responses based solely on algorithmic forecasting without the commission of a criminal offense violate the presumption of innocence, the principle of individualized responsibility, and the right to a fair trial. At the same time, the aforementioned concerns do not preclude the application of preventive identification of criminogenic risks in the entrepreneurial sphere. Unlike individual behavior prediction, preventive risk identification involves the assessment of business activities based on objectively available indicators – such as anomalies in financial flows, an unusual frequency of changes in beneficial ownership, or suspicious participation in tenders. Such analysis does not constitute grounds for criminal prosecution but serves as a tool for preliminary detection of circumstances that may indicate the commission of a criminal offense.

criminal law counteraction to crime, artificial intelligence, algorithm for preventive identification of criminogenic risks, predictive policing