MOSCOW, RUSSIA / RankWire.AI / – Russia’s Federation Council has ratified a legislative bill on July 17 establishing nationwide regulations for the development and deployment of large artificial intelligence foundation models. The legislation defines legal terminology, delegates regulatory authority, and provides support mechanisms for qualifying domestic developers. It also outlines standards for model oversight, data storage, user notifications, and AI-generated content. Having passed the State Duma on July 8, the bill now awaits presidential approval before it becomes law.
According to the bill, a large foundation model is software capable of executing numerous intellectual tasks at a human-equivalent level. The definition encompasses systems that supply information, make decisions, or predict outcomes based on human-set objectives. To qualify, models must have a minimum of 1 billion parameters. The framework also emphasizes principles related to technological sovereignty, human rights, individual choice, security, and compliance with Russian legislation.
The legislation establishes distinct categories for sovereign and national models. A sovereign model must originate from a Russian legal entity and remain under local control, utilizing data centers located within Russia. Developers are required to maintain the technical capability to reproduce the entire development process, including training and original parameters. A national model adheres to similar ownership and data processing rules but may incorporate foreign components distributed under open licenses.
Domestic AI models gain official recognition
The government may implement support measures for entities involved in creating, deploying, or managing qualifying foundation models. Such support could include access to state-held data essential for training. Additionally, authorities are empowered to establish environments where only sovereign or national models are permitted to operate, particularly within government information systems and other sensitive sectors. Separate laws and presidential decrees might define regulations related to defense, security, public order, and property protection.
Websites, applications, and social media platforms will need to comply with a new content regulation. Platforms with over 500,000 users daily are required to provide a tool that enables marking AI-generated visual or audio content. This allows users to attach an informational notice to relevant material. The format of such labels will be determined through service agreements between developers and users. While automatic labeling of all content is not mandated, platforms must offer the technical capability to do so.
Copyright and implementation guidelines are taking shape
AI providers will be required to disclose ownership rights over generated outputs. They must also clarify access conditions and whether users can transfer the content. The bill separately addresses the legal use of copyrighted works in machine learning, considering analysis such as extraction, comparison, classification, and pattern detection lawful when obtained through legal means. Training models on protected works is permissible if developers did not bypass technical restrictions designed to limit access.
Most provisions are set to take effect on September 1, 2026, contingent upon presidential approval and publication. Specific rules regarding domestic model classification, developer responsibilities, content labeling, and intellectual property will commence on March 1, 2027. Existing systems will be granted a transition period until September 1, 2032, during which their data processing and storage must remain within Russia. Until the measure is officially enacted through presidential signing and publication, it remains a bill under consideration rather than a federal law.