In a major move to solidify its “digital sovereignty,” the Russian government is preparing a new legislative framework that could effectively ban or strictly restrict foreign AI models—including ChatGPT, Gemini, and Claude—by September 1, 2027.
The proposal, published by the Ministry for Digital Development on March 20, 2026, targets “cross-border artificial intelligence technologies” that transmit Russian user data to foreign servers.
The 2027 “Data Lockdown”
The core of the new regulation is a strict data localization requirement. For a foreign AI service to remain legal in Russia, it must meet the following criteria:
- Local Storage: All user queries, chat histories, dialogues, and metadata must be stored on physical servers located within Russian territory.
- 3-Year Retention: This data must be kept for a minimum of three years and be accessible to Russian authorities upon request.
- Audience Threshold: The rules apply to any AI platform with more than 500,000 daily active users in Russia.
- Content Filtering: Models must be adjusted to respect “traditional Russian spiritual and moral values” and prevent “covert manipulation” by foreign algorithms.
The “Sovereign AI” Push
The Ministry stated the rules are designed to protect citizens from “discriminatory algorithms” and the “unauthorized transfer of sensitive personal data” to the U.S. and its allies.
| Category | Impact of 2027 Law |
| Western Models | Likely to be blocked as companies like OpenAI and Google historically refuse local data storage demands. |
| Domestic Tools | Expected to benefit Sberbank (GigaChat) and Yandex (YandexGPT), which already comply with local laws. |
| Chinese Models | Open-source models like Qwen (Alibaba) or DeepSeek may be permitted if deployed on Russian-owned infrastructure. |
A “Digital Prison” or “Digital Sovereignty”?
The move has sparked a fierce debate over the future of the Russian internet:
- The Government View: Ministry officials argue the law is essential to prevent Western AI from being used as a tool for “information warfare” and to build a self-sufficient domestic AI ecosystem.
- The Critic’s View: Human rights groups and digital activists, including the Center for Countering Disinformation (CCD), have labeled the initiative a “digital prison,” noting that it coincides with a 2.27 billion ruble investment into AI-powered internet traffic censorship systems set for 2026.
- The “API” Loophole: Experts warn the law will most heavily impact Russian developers who use APIs to build local bots. While many Western services are already “blocked” via IP, this law would criminalize the reselling or integration of these tools into local apps.
Implementation Timeline
- 2026: Final government review and approval of the “Cross-Border AI” bill.
- Late 2026: Launch of Roskomnadzor’s AI-powered traffic filtering system to detect unauthorized AI usage.
- September 1, 2027: The law officially enters into force; non-compliant foreign models face a total ban.
