ALGORITHMIC TRANSPARENCY AS A PRINCIPLE OF LEGAL REGULATION OF ARTIFICIAL INTELLIGENCE

The article examines the issue of algorithmic transparency as a key principle in regulating artificial intelligence in the context of digital transformation of public administration. It substantiates the need to move from formal disclosure of AI system logic to the creation of mechanisms that ensure real interpretability and accountability of decisions made. It is noted that both public authorities and citizens are interested in ensuring AI transparency, as access to decision explanations forms the basis for legal protection and trust in digital services. Special attention is given to the concept of counterfactual explanations as a means of achieving a balance between technical feasibility and legal verifiability of algorithmic decisions. The article analyzes ethical, legal, and social risks associated with the «black box» nature of AI, including the threat of algorithmic bias and discrimination. It also reviews strategic documents of the Russian Federation related to explainable AI, as well as foreign approaches to interpretability. The purpose of the study is to justify algorithmic transparency as a fundamental condition for the legal regulation of AI and to propose directions for its institutional implementation. The methodological basis of the study includes general scientific methods (analysis, synthesis, induction, deduction), as well as specialized legal methods: formal legal analysis of regulatory acts and comparative legal analysis of international approaches to explainable AI. The study concludes that a multidisciplinary approach, encompassing technical, legal, and ethical elements, is necessary to ensure AI accountability and to protect the rights of individuals in the digital legal space.