Модели машинного обучения: в поисках места в системе объектов интеллектуальной собственности

Закон. 2026. № 1. С. 70–77.

Savelyev A.

This article explores the rationale for establishing a separate legal regime for machine learning models as a distinct intellectual property object, distinct from the legal regime for computer programs, under which they are classified under current Russian legislation. This paper provides a structural analysis of machine learning models, including an examination of their architecture, parameters, weights, and program code, revealing specific characteristics not found in traditional software. The author argues that the functional significance of machine learning models in artificial intelligence systems, particularly in AI assistants and AI agents endowed with the ability to make legally significant decisions, necessitates the development of specialised legal mechanisms for regulating their quality, transparency, version control, and protection from unauthorised copying and external influences. The article argues that the existing legal regime for computer programs is insufficient to adequately regulate legal relations related to artificial intelligence models. The article substantiates the position that the introduction of a special legal regime for machine learning models will create a regulatory framework for the systemic integration of artificial intelligence provisions into the Civil Code and ensure legal support for the implementation of relevant technologies.

Keywords: программа для ЭВМ, computer software, Commercial AI models, параметры модели ИИ, модель ИИ