Ways to minimize the risks of introducing AI agents through public administration mechanisms

Background. The widespread introduction of artificial intelligence technologies has led to an awareness of the risks, the validity of which requires research into the mechanisms of AI operation, as well as the consequences for all aspects of social life and the possibilities of public administration. The lack of transparency for users and consumers regarding the mechanics of AI agents as autonomous systems that solve specific tasks, combined with the comfort provided by the automation of many processes that are already beyond the scope of traditional operations, increases the risk associated with the widespread use of such technologies. The validity of these concerns and the state's position on these risks have been assessed based on an analysis of the principles of AI agents' data handling. The article highlights specific risks associated with the implementation of AI agents related to information processing technology. The purpose is to find ways to minimize the risks of introducing AI agents into social life and the state economy by analyzing the mechanisms of operation of individual autonomous intelligent systems. Methodology. Abstract logic was used as the main method, and secondary analysis of third‑party research was also used. Results. The risk factors for the functioning of autonomous intelligent systems are summarized, and the directions of state regulation in the field of artificial intelligence technologies are clarified. Practical implications. The results obtained can be used in the development of government strategies for regulating AI and in the formulation of regulations.