Russia's Sovereign AI Plan Faces Technical Reality Check From Industry
Russia's largest oil company, Rosneft, has told the government that its draft "sovereign AI" law won't work as written. The company, led by Igor Sechin, a close Putin ally, said building AI systems in isolation from global technology is technically unworkable.
The draft bill, prepared by Russia's Ministry of Digital Development, requires neural networks to be developed and trained only by domestic companies using Russian datasets. Rosneft countered that Russia lacks the computational infrastructure and sufficient Russian-language datasets to meet those requirements.
The company proposed allowing publicly available internet data regardless of server location-including Wikipedia-for AI development. Other industry groups echoed the concern: the Digital Platforms Association, the Association of European Businesses, and the Chamber of Commerce all said no Russian AI models currently meet the bill's "sovereignty" criteria.
The Definitional Problem
Industry groups also flagged confusion in the bill itself. The Association of Computer and Information Technology Enterprises noted that "sovereign" and "national" AI models have identical basic requirements, with specifics still undefined.
Business leaders objected to another provision: requiring only "trusted" models that comply with government safety standards and process data exclusively within Russia's borders. The Association of European Businesses criticized the inclusion of "traditional Russian spiritual and moral values" as a criterion, arguing that concepts like "high moral ideals" and "the priority of the spiritual over the material" aren't legal categories and shouldn't guide AI approval decisions.
What's at Stake
Industry representatives warned that passage in its current form would increase implementation costs, delay product launches, and push AI development projects to other countries. Russian citizens could also face limited access to advanced diagnostic and medical technologies.
The draft bill is part of a broader push. In March, the Ministry for Digital Development proposed regulations to limit or ban foreign ChatGPT, Claude, and Gemini. The stated goal is reducing "covert manipulation" and discriminatory algorithms while promoting domestic alternatives like Sberbank and Yandex.
Generative AI and LLM development in Russia would face new restrictions under these rules.
Separate Issue: Disinformation Networks
In a separate development, OpenAI said it dismantled accounts linked to a Russian disinformation project called "Rybar." The accounts were generating content in Russian, English, and Spanish for coordinated campaigns, with material distributed through affiliated accounts.
Your membership also unlocks:
