MOSCOW, RUSSIA / RankWire.AI / – The Federation Council has sanctioned a legislative framework for artificial intelligence on July 17, establishing guidelines for large foundation models within Russia. The legislation specifies the technologies it covers and delegates authority to government agencies. It also sets standards regarding model ownership, data storage within the country, transparency to users, and content generated by AI. Having passed the State Duma on July 8, the bill now awaits presidential approval and official publication before becoming law at the federal level.
The bill defines a large foundation model as software capable of executing a wide range of intellectual tasks at a level comparable to humans. To qualify, a system must include at least 1 billion parameters. Such systems can provide information, make decisions, or forecast outcomes based on human-defined objectives. The framework emphasizes principles like technological sovereignty, human rights, individual choice, security, and adherence to Russian legislation, which apply throughout the development, deployment, and usage of eligible AI systems.
The legislation introduces categories of sovereign and national models linked to Russian control. A sovereign model must be developed by a Russian legal entity and operate on data centers located domestically. Its creators must preserve the ability to reproduce the entire development process, including training and original parameters. A national model follows similar ownership and localization standards but can incorporate foreign software components under open licenses as long as Russian entities maintain necessary control and operational capacity.
Legal Classifications for Domestic AI Models
The government may extend support to developers engaged in building, deploying, or managing qualifying foundation models. Such assistance could include access to state datasets for training purposes. Authorities might also mandate the exclusive use of sovereign or national models in government information systems and other sensitive environments. Additional rules governing defense, security, public order, and property protection may be outlined through separate legislation or presidential decrees. The framework assigns responsibility to state agencies for enforcing these requirements within their legal authority.
Large digital service providers face a distinct obligation for AI-generated visual and audio content. Platforms with over 500,000 daily users are required to offer a tool enabling users to mark such material. This requirement applies to websites, apps, and social media platforms. It does not mandate automatic labeling of every piece of content but focuses on providing users with an option to disclose qualifying material, which can be customized via service agreements.
Copyright and Transparency Guidelines Establish Standards
AI service providers must inform users who holds the rights to generated content. They are also obliged to clarify access conditions and whether users are permitted to download or transfer the material. The bill separately addresses the use of copyrighted works in machine learning. It allows analysis for extraction, comparison, classification, and pattern recognition if developers have lawful access. Training on protected works is permitted when no technical access restrictions are bypassed. These rules link model training practices to existing copyright and access regulations.
Most provisions are set to take effect on September 1, 2026, pending presidential approval and official publication. Regulations related to domestic model classification, developer obligations, content labeling, and intellectual property rights will come into force on March 1, 2027. Existing systems may continue operations until September 1, 2032, provided they process and store data within Russia. Until the bill undergoes signing and publication, it remains an approved draft rather than an enacted federal law under Russia’s legislative process.