Algorithmic Discrimination as a Socially Dangerous Act: the Concept, Signs and Prospects of Criminalization

Objective: to define the concept and essence of algorithmic discrimination from the perspective of criminal law science and to substantiate the need to rethink this phenomenon in criminal law as artificial intelligence is introduced in all spheres of society and the state.

Methods: the research methodology consists of a set of complementary methods of scientific cognition. The comparative-legal analysis was used to study foreign criminal legislation and doctrinal approaches to the qualification of algorithmic discrimination. The formal-legal method ensured accuracy in the interpretation of the current norms of the Russian criminal law. The systematic and logical-legal analysis revealed the cause-and-effect relationships between the architecture of artificial intelligence systems, the nature of training data and the socially dangerous consequences of their use. The doctrinal analysis provided conceptual continuity with the established international legal approaches to studying discrimination.

Results: it was established that algorithmic discrimination is not a fundamentally new phenomenon, but a technologically mediated continuation of historically harmful stereotypes and biases. The research identified its distinctive features compared to classical discrimination forms: the massive and systemic nature of the harm caused, the lack of transparency of cause-and-effect relationships, the blurring of the subject matter, and the criminal-legal significance of the careless guilt. The existing administrative and legal protection mechanisms were proved to be insufficient to counteract this phenomenon in the context of digital public administration, and therefore the need to criminalize algorithmic discrimination as an independent crime was confirmed.

Scientific novelty: the work is a systematic study of algorithmic discrimination from the perspective of criminal law. The criminal-law definition of this phenomenon was formulated; its essence was revealed through the causal relationship between historically established bias, the use of artificial intelligence systems and socially dangerous consequences. The specific features of the corpus delicti that are uncharacteristic of traditional forms of discrimination were identified. Among them, the article substantiates the criminal-legal significance of the careless guilt and establishes the role of the state as a potential subject of responsibility for introducing discriminatory algorithmic systems into public administration.

Practical significance: the study results are important for application in both law making and law enforcement activities. In the field of legislation, they can be used to improve the corpus delicti provided for in Article 136 of the Russian Criminal Code titled “Violation of equality of human and civil rights and freedoms”, or to criminalize algorithmic discrimination as an independent component. In the field of law enforcement, the proposed approaches allow more accurately assessing the socially dangerous consequences of using artificial intelligence systems, qualifying the subjective side of the relevant acts, and developing criminal-legal mechanisms to prevent the spread of discriminatory practices in digital governance.