⚖️ Expert outlines upcoming AI regulation in Russia

The upcoming legislation is designed to transition the industry from voluntary ethical codes to a binding legal framework, codifying fundamental legal terminology and establishing liability guidelines for all parties involved. Alexander Zhuravlev, Managing Partner of EBR Law Firm and an active participant in drafting the regulations, detailed the scope of the framework in an interview with the Kommersant newspaper.

International insights and domestic priorities

While drafting the bill, lawmakers closely analyzed global regulatory frameworks; however, rather than copying foreign policies verbatim, they adapted them to serve national interests. This intersection of global approaches has led to a distinct regulatory model.

• Risk-Based Approach (EU Precedent): Similar to the European Union’s AI Act, the Russian framework proposes categorizing AI systems based on risk tiers and introducing stringent transparency requirements. However, the specific metrics for these classifications have not yet been formalized in the current draft.
• Mandatory Labeling (US and China Precedents): The legislation introduces a requirement to label AI-generated content, aligning the domestic model with American and Chinese initiatives aimed at combating misinformation.
• Core Priorities: Unlike the European model, which prioritizes ethics and individual human rights, the Russian approach places primary emphasis on technological sovereignty and national security. For the public sector, this introduces the concept of “trusted models,” which will be subject to security screening by the FSB and FSTEC.

Critical debates and regulatory risks

Despite the clear necessity of a legal framework, the current draft has sparked intense debate across the business community and the creative industries.

The most contentious issue remains the uncompensated training of domestic AI models on proprietary commercial datasets. The current draft permits data mining of copyrighted works without classifying it as intellectual property infringement. Legal experts warn that this undermines the core principles of the Civil Code of the Russian Federation by allowing developers to derive commercial value from millions of third-party works without providing financial remuneration to authors.

Furthermore, the ambiguous definition of “original creations” produced by AI systems remains heavily disputed. This concept directly conflicts with the Civil Code, which traditionally recognizes only human beings as authors.

The proposed liability mechanism introduces additional pressure on the market: developers and operators will effectively bear the burden of proof to demonstrate their lack of culpability if an AI system generates unlawful outputs. Industry insiders warn this shift could disincentivize local innovation and product development.

Industry incentives and implementation timeline

To balance the strict regulatory requirements, lawmakers have embedded support measures for the technology sector into the draft. Data centers are expected to receive subsidized electricity tariffs, streamlined infrastructure connectivity, and tax preferences.

The legislation is scheduled to take effect on September 1, 2027. The authors describe this extended transition period as a deliberate measure to grant the AI market sufficient time to evaluate regulatory risks, adapt product pipelines, and draft necessary sub-legislative acts without stifling a highly dynamic IT sector.