MMOOZZGG.com

Basics

AI and personal data law in Russia: what you can give a model and why a local setup helps

3 min read12 sources

AI and personal data law in Russia are simpler for a local setup than they seem: personal data needs a legal basis and a place of storage, the large foreign APIs are officially unavailable, and a 2026 law introduces model statuses. Below, point by point; this is not a legal opinion.

On this page · 4 sections

Personal data

Personal data is not only a name and a phone number: an individual’s tax number and data from open registers count too. Sending them to a foreign service raises the question of cross-border transfer. The main problem, according to the author of a review the publication cites, is that most international providers of language models have no processing contract. Replacing names with tokens such as user_123 does not help if the link can be restored. Relatively safe options, according to the same publication: Russian cloud platforms with a contract, a model inside your own infrastructure, or anonymization before sending.

Access to foreign APIs

As of 29 September 2026, Russia is not on the list of supported countries of two large foreign providers. Since 9 July 2024 one of them has blocked API traffic from unsupported countries, and under its rules, access from unsupported countries, including through intermediaries, may lead to an account block. For a business this is an operational risk, not only a legal one: the channel can disappear any day.

Law 243-FZ

Since 1 September 2026 Russia has had a law on support for AI technology development: statuses of a sovereign and a national model, criteria from 1 March 2027, and a transition period for existing systems until 2032. From 1 March 2027 the government will be able to define cases where only such models are allowed. Our blog has an analysis for business.

Three classes of data

Public and anonymized data: any API. Personal and commercial data: a Russian cloud with a processing contract, or a local model. Source code, medical and credit documents, state secrets and critical information infrastructure: only a local setup. This is what a telecom operator, Moscow healthcare and an airport from this section’s cases did, and a developer chose an isolated setup on a Russian model. After source code leaked through a chatbot in 2023, a large electronics maker banned outside services and began building its own.

Questions and answers

Can you use a foreign model through a Russian intermediary?

The data still goes abroad, you have no processing contract with the provider, and the provider itself warns that it may block accounts accessed from unsupported countries. The risk stays with you.

Do you need to notify Roskomnadzor, Russia’s data protection regulator, about a local model?

A local model does not change your duties as a data operator; it removes the cross-border transfer and third parties. Your notice of personal data processing stays as it was.

Sources and fact-checking

Checked by the editors on 29 September 2026.

  1. Anti-Malware, 152-FZ and neural networks, 17 June 2026 (anti-malware.ru)
  2. Claude Platform, supported regions (checked 29 September 2026) (platform.claude.com)
  3. OpenAI, supported countries for the API (checked 29 September 2026) (developers.openai.com)
  4. Caixin, OpenAI blocks API traffic from unsupported countries from 9 July 2024, 26 June 2024 (caixinglobal.com)
  5. CNews, which AI models will get sovereign status, 28 September 2026, material marked as advertising (corp.cnews.ru)
  6. Tech Monitor, Samsung banned ChatGPT after a leak, 2 May 2023 (techmonitor.ai)
  7. Digital Platforms, case no. 3570, 2 September 2026: Ministry of Justice of Russia (platforms.su)
  8. Digital Platforms, case no. 3509, 2 September 2026: Moscow Department of Health (platforms.su)
  9. Digital Platforms, case no. 3148, February 2026: T2 (Tele2) (platforms.su)
  10. ETH Zurich, Apertus press release, 2 September 2025: ETH Zurich, EPFL and CSCS, Switzerland (ethz.ch)
  11. Federal Law No. 243-FZ of 26 July 2026, text on ConsultantPlus (consultant.ru)
  12. Federal Law No. 152-FZ “On Personal Data”, text on ConsultantPlus (consultant.ru)

Found a mistake? Tell us

Publications

Cases on this topic: 5

  • Market case · Russia

    Ministry of Justice of Russia AI lawyer avatar with a base of federal laws; a local language model is being trained for it

    The country’s first AI lawyer avatar understands spoken language and advises on legal questions; its knowledge base holds federal laws, regulations and information from the ministry’s website. As of July 2026, specialists are training a local language model, which runs on the organization’s own servers, to speed up the work and improve the answers.

    Government

    2 September 2026 · source: platforms.su

    Read the case
  • Market case · Russia

    Moscow City Health Department AI assistant for hospital analytics on a local language model: data does not leave the network

    An AI assistant built on a locally deployed large language model sits inside the data marts of the city’s unified medical system. Local deployment was chosen to work with sensitive medical data without sending it outside.

    Government

    2 September 2026 · source: platforms.su

    Read the case
  • Market case · Russia

    T2 (Tele2) tender for a self-hosted AI coding assistant in a closed data center: 15 developers, response within 400 ms

    In February 2026 the operator announced a tender for a self-hosted AI coding assistant that runs in the company’s closed data center in Rostov-on-Don, so source code never leaves its infrastructure. Requirements: 15 users, autocomplete with a response of at most 400 ms, six programming languages.

    Telecom

    2026 · source: platforms.su

    Read the case
  • Market case · World

    Samsung Electronics ban on outside generative AI services after a source code leak, own tools in development

    The Samsung ChatGPT ban came in May 2023: the company stopped employees from using outside generative AI services on work devices after three employees uploaded source code and notes from internal meetings to a chatbot. An internal memo warned of disciplinary action, up to dismissal.

    World

    2 May 2023 · source: techmonitor.ai

    Read the case
  • Market case · World

    ETH Zurich, EPFL and CSCS, Switzerland fully open sovereign model Apertus: 8 and 70 billion parameters, 15 trillion tokens, more than 1,000 languages

    Two federal institutes of technology and the national supercomputing center released a fully open language model in two sizes, 8 and 70 billion parameters. It was trained on 15 trillion tokens in more than 1,000 languages, and 40% of the data is not in English. More than 10 million GPU hours on the Alps supercomputer went into the model and related research.

    World

    2 September 2025 · source: ethz.ch

    Read the case
Contact us