Abstract
The article is devoted to a theoretical and legal analysis of the transformation of the fundamental principles of legal responsibility under the influence of autonomous artificial intelligence systems. The relevance of the study is обусловлена a qualitative transformation of social reality, in which algorithmic systems increasingly act not merely as passive tools but as active elements of legal relations, capable of independently making decisions under conditions of uncertainty. The paper substantiates that the preservation of an anthropocentric model of responsibility in the context of the de facto autonomy of algorithms gives rise to a systemic contradiction: classical legal institutions of fault, causation, and the subject of an offense cease to adequately describe the mechanism of harm causation. The author identifies three levels of transformation of the legal construct of responsibility: (1) a change in the structure of the subject composition (the emergence of “algorithmic intermediation”); (2) a deformation of the causal link (the multi-layered, probabilistic, and opaque nature of algorithmic decision-making); (3) a blurring of the criteria of fault (the uncertainty of the boundaries between intent, negligence, and a “technologically inevitable error”). Proposals are formulated for the evolutionary adaptation of the institution of legal responsibility to the realities of the digital age, including the introduction of the principle of technological neutrality, the establishment of standards for algorithmic transparency, and the implementation of differentiated liability regimes depending on the degree of system autonomy. The author concludes that technological development does not eliminate the need for legal responsibility as such, but requires its conceptual rethinking: responsibility remains inherently human in nature; however, the mechanisms of its implementation must be adequate to the new technological reality.