Russia bets on domestic AI but won’t ban foreign models

The Russian government has greenlighted a new bill designed to regulate and support the development of artificial intelligence in the country.

The legislation, which aims to promote domestic AI solutions, will not ban the use of other neural networks or the training of local models with foreign data.

Russia sets out to regulate artificial intelligence

A government commission overseeing legislative initiatives has approved a draft law that should facilitate the development of artificial intelligence (AI) in Russia.

The bill was given the nod on Monday, according to a report by the Interfax news agency quoting the press office of Deputy Prime Minister Dmitry Grigorenko.

The document is dealing with “large fundamental models with more than 1 billion parameters,” the announcement highlighted.

These have been divided into two main categories – “sovereign”, exclusively developed by Russian entities using local infrastructure, and “national,” which may partially rely on open-source components but must again be largely developed domestically.

An earlier version of the legislation featured a third category – “trusted” AI models – covering those designed to work with critical information and infrastructure, as already reported by Cryptopolitan.

The latter has now been dropped and Grigorenko explained that the proposal has been revised taking into account the suggestions of businesses.

The main focus remains unchanged – supporting Russian-made AI models, the official indicated, adding that the executive power is currently considering various incentives.

The bill creates a regulatory framework for the development of AI in Russia and provides key legal definitions such as “artificial intelligence” and “large language model,” he said and elaborated further:

While the key goal of the law is to stimulate the development and demand for domestic solutions, it does not prohibit the use of foreign neural networks.

Dmitry Grigorenko made that clear in an earlier statement, when he also emphasized that the nation’s public sector will primarily utilize Russian-built models.

Russian AI law to enter into force by the fall

Russia does not currently have a legal framework for artificial intelligence, which it is already developing and implementing, the deputy prime minister noted.

Laws and rules are needed, however, to make any significant decisions regarding the new technology and introduce support and protection mechanisms, Grigorenko insisted.

The updated bill may be submitted to the State Duma, the lower house of parliament in Moscow, by the end of the week, a knowledgeable source told the business daily Kommersant.

The main provisions of the future law “On Supporting the Development of Artificial Intelligence Technologies in the Russian Federation” are expected to enter into force on September 1, 2026, the newspaper unveiled.

Other texts, such as those determining the powers of government agencies and the responsibilities of developing entities, will come into force in March 2027.

The authors have significantly simplified and shortened the document, bringing it down to just 13 articles and pages.

One of the original regulations that has been scrapped is the obligation for developers to label all AI-generated content as such. The new requirement is only to provide users with this option.

Various proposals to hold AI service providers accountable for misuse of their technology have been reduced to the vague “shall be liable in accordance with Russian law.”

Texts regulating the use of copyrighted materials to train AI, as well as restrictions for so-called “cross-border” AI models, have been deleted as well.

At the same time, Russian authorities are leaving the door open to enacting additional regulations in the future that may include some of the now retracted provisions, like the rules for AI data centers.

If you're reading this, you’re already ahead. Stay there with our newsletter.