AI models developed in Russia will be divided into categories, depending on their reliance on domestic or foreign solutions, according to a new bill drafted in Moscow.
While the legislation is being finalized by the government, Putin is already taking the reins of artificial intelligence by setting up a Kremlin commission that will oversee its implementation.
Russia drafts law to regulate artificial intelligence
The authorities in Moscow are preparing to classify artificial intelligence (AI) models into two or three groups based on how Russian they are by design.
According to the current version of the government bill “On Artificial Intelligence in the Russian Federation,” they will be categorized as “sovereign,” “national” or “trusted.”
Each model will be approved through certification upon meeting the requirements of the Federal Service for Technical and Export Control (FSTEK) and the Federal Security Service (FSB).
Officials want software providing access to Russian AI pre-installed on every smartphone and tablet, the business daily Kommersant reported after examining a copy of the document.
To be labeled as “sovereign,” an AI model must be developed and operated in Russia, by its citizens and Russian companies.
It will also have to be trained exclusively using datasets generated within the Russian Federation, and without any components built outside the country.
Employing foreign open-source solutions and using non-Russian datasets for training will be permitted in the case of “national” artificial intelligence models.
The third category of “trusted” AI models will cover those designed to work with critical information and infrastructure, the article detailed further.
They will be evaluated against a set of government-elaborated criteria, and their security must be confirmed by the FSTEK and the FSB.
Experts are skeptical about ‘sovereign’ AI
While the bill is still under review and may undergo revisions before it’s adopted, Moscow is certainly obsessed with “AI sovereignty” and is likely to continue to push in that direction.
Representatives of the industry warn, however, that creating an AI model from scratch, without using foreign open-source solutions, would be expensive and impractical.
Fully sovereign platforms are virtually nonexistent on the market today, a source from a leading Russian IT company told Kommersant, opting to remain anonymous.
Even the largest tech firms are using open-source tools and libraries when developing their own models, noted Just AI co-founder Kirill Petrov.
In Russia, only Sberbank, the majority state-owned financial giant, is working on a fully independent model. Last fall, it unveiled the nation’s first AI-powered humanoid.
According to MWS AI, building a local alternative to foreign models would require hundreds of billions of rubles, and the costs will inevitably be passed onto users.
It would be a waste of money to join the race for AI models, instead of investing in world-class open-source products, added the company, which specializes in generative AI solutions.
Putin creates Kremlin commission on AI development
Meanwhile, Russian President Vladimir Putin signed a decree this week for the establishment of a “presidential commission on the development of AI technologies.”
The move comes after Putin urged Russians to rally behind domestic artificial intelligence and tech sovereignty in November, announcing the upcoming creation of a national AI task force.
At the time, he emphasized that Russia could not allow itself to become dependent on foreign neural networks, as this is a matter of sovereignty.
The body will be co-chaired by Deputy Prime Minister Dmitry Grigorenko and the Deputy Chief of Staff of the Presidential Executive Office Maxim Oreshkin.
The ministers of defense and digital development, as well as Sberbank’s chief executive and the president’s special envoy for tech development, will be sitting on it, too, as reported by the business news portal RBC on Thursday.
The commission will primarily focus on coordinating the efforts of federal, regional, financial, and other authorities toward introducing AI technologies in various fields.
It will also produce a strategy for AI implementation, facilitate the adaptation of economic, social, and public sectors to artificial intelligence, and propose initiatives for international AI cooperation.
