Russian developers will be able to include software created with the help of neural networks in the domestic software registry without separate restrictions. This eliminates the risk of rejection for businesses due to AI code, but the costs of tokens and responsibility for the security of the final product remain with the companies themselves.
The decision affects virtually the entire industry. According to estimates by the Ministry of Digital Development, the share of developer companies using vibe coding — creating programs from text prompts with the help of AI — has already reached 90%. For customers and users, something else matters more: a significant portion of such development is carried out on foreign neural networks, so control over the final code remains the key task.
A registry without barriers, but under control
Minister of Digital Development Maksut Shadayev said on November 12 that there will be no separate restrictions for software created with the help of generative AI when it is included in the registry. According to him, such practices are being actively adopted, including by large IT companies, but the ministry is not yet recording a sharp increase in the number of applications.
The minister attributed this to the fact that the technology is at the stage of adoption: companies are mainly creating prototypes and solving individual tasks rather than releasing large products capable of replacing entire classes of corporate software. According to Shadayev's estimate, Russia, which had to replace a large volume of foreign software, could become one of the leading countries in generative development. At the same time, businesses are advised to watch out for two risks: the real costs of tokens (the units in which the operation of neural networks is paid for) and product security.
Shadayev explained that code is written in external services and open models, while Russian models are still used infrequently. "Our task is to ensure that the final code is still controlled, even if part of it was generated. So that a specific company is responsible for the final product and can develop and implement it," the minister emphasized.
He estimated Russia's lag behind global leaders at 6–12 months: its own competencies are needed in case open models are closed, but as long as they are available and in some cases more effective than Russian ones, the country does not limit itself in using them.
Subsidies, AI agents and security
Access to the registry is also important for receiving state support: two months ago it became known that ministries had preliminarily agreed on subsidies for developers who give small and medium-sized businesses discounts on AI products included in the registry. The state compensates up to 100% of the cost in the first year, up to 50% in the second and up to 25% in the third (no more than RUB 200,000 per year per client), and model creators were offered a regional tax deduction.
The use of AI is already going beyond development: Sberbank CEO German Gref said that the main competitors of marketplaces will not be other platforms but buyers' personal AI agents, and the new business model will abandon the commissions on which the platforms currently earn. Kazakhstan President Kassym-Jomart Tokayev, at the CSTO summit in Moscow, called for joining efforts on information security given the growing influence of AI.
Cryptalist analysis: Allowing AI code into the domestic software registry is an institutional signal that generative development is moving from the experimental into the legitimate commercial sphere. Removing barriers stimulates growth in the number of applications and accelerates import substitution, but shifting responsibility for security and token costs onto businesses creates a new class of operational risks. For the market, this means increased competition among domestic developers and higher demand for tools for auditing and controlling generated code.