Russia’s Upper House Greenlights Landmark AI Regulation Law

The Federation Council has signed off on a new law that sets the rules for developing, rolling out, and overseeing artificial intelligence in Russia. The legislation, approved on July 17, creates a legal framework for managing large foundational AI models—programs with at least one billion parameters that can match or exceed human-level performance in tasks like software creation.

Under the law, two distinct categories of these models emerge: sovereign and national. A sovereign model must be built by a Russian legal entity that retains full control at every stage of its lifecycle, from design to deployment, and must be hosted in domestic data centers. A national model, by contrast, can incorporate foreign components as long as they’re used under an open license. Both types must comply with Russian law and traditional values.

Developers working on these models will qualify for government support, including financial aid, property assistance, guarantees, and access to information. They’ll also be allowed to train neural networks on data from federal and regional government information systems. The exact terms for accessing that data will be set by the government in coordination with security agencies.