VULNERABILITIES OF ARTIFICIAL INTELLIGENCE SYSTEMS USED IN ASSETS OF INFORMATION AND TELECOMMUNICATION NETWORKS OF ORGANIZATIONS

Recently, in the information sphere of Russia, including objects of Critical Information Infrastructure (CII) (these include Information and Telecommunication Networks (ITCNs) of organizations-subjects of CII), Artificial Intelligence (AI) systems are being used more and more actively and widely. The introduction of AI systems into CII objects requires special attention to ensuring their security, reliability, and operational resilience, since AI technologies can both increase the efficiency and automation of processes in CII objects and simultaneously create new risks and vulnerabilities. This requires an integrated approach to managing them. The relevance of research and search for solutions in this area is also emphasized by the preparation of the GOST R draft entitled "Artificial Intelligence in Critical Information Infrastructure. General Provisions." In this regard, the article’s goal is to systematize the input data obtained from various sources to monitor vulnerabilities and assess their applicability to ITCN assets as part of the vulnerability management process, as an organization's activity to analyse and eliminate vulnerabilities identified in various ITCN assets. The article introduces the definition of AI systems and their classification. Furthermore, the AI systems' potential use for ensuring the information security of organizations' ITCN assets is proposed. The vulnerabilities of AI systems are summarized on the basis of various analytical studies’ results, and their examples are given. A selection of useful Internet information sources on the vulnerabilities of AI systems is provided. In conclusion, the research results obtained are summarized, and some of the recommendations increasing the security of AI systems used by organizations in СII objects are given.