Legal Mechanisms of AI Assessment in China and Russia: A Comparative Study

Abstract

This article presents a comparative analysis of the legal mechanisms of AI assessment in China and Russia. The author employs the functional and dogmatic methods, together with a basic concept—the “AI assessment mechanism”—as the foundation for comparing heterogeneous legal institutions. The comparison is conducted across seven parameters: the object and subject matter of assessment, the criteria for selecting the object, the procedure, the timing, the legal force, and the regulatory purpose. In China, the mechanism is based on algorithm registration and security assessment, established by the Cyberspace Administration of China acts and implemented through administrative practice and technical standards. In Russia, the first legally significant assessment mechanism is introduced by Federal Law No. 243-FZ of July 26, 2026, which establishes a register of large foundation models and a requirement to confirm their compliance with legislation and traditional Russian spiritual and moral values. The comparison reveals two types of relationship: divergence and convergence. The divergence stems from the regulators’ different goals: China develops assessment as an instrument of control over the information space, while Russia designs it as a tool for securing technological sovereignty. There is convergence, however, in the choice of instruments: both legal orders rely on registration or record-keeping and on security or compliance assessment, and both mechanisms operate on an ex ante basis. The article and its findings represent the first comparative study of these mechanisms. The Chinese experience is particularly relevant to the unresolved issues Russia faces in designing procedures for registering large foundation models and confirming their compliance with applicable requirements.