Legal Framework for AI: Balancing Development and Control

The event that took place on July 7 reflects the authorities’ desire to prevent the country from falling behind technologically while at the same time creating a manageable trajectory for the integration of AI into the economy. At the same time, the purpose of this document is not to resolve all contentious issues at once, but to create a legal foundation that allows for more detailed regulation through secondary legislation. Such an approach, on the one hand, provides flexibility, and on the other hand, generates uncertainty that may become a barrier to long-term investment and strategic planning in the industry.

The essence of the initiative is revealed in the very logic of its presentation: the draft law was introduced by Deputy Minister of Digital Development Ivan Lebedev with the participation of Minister Maksut Shadayev and Deputy Prime Minister Dmitry Grigorenko. The emphasis was placed specifically on stimulating technological development rather than on strict regulation. According to Lebedev, the structure of the document is the result of extensive discussions with businesses and government agencies and is intended to ensure balanced AI development in the Russian Federation. From an academic perspective, such a compromise approach can be interpreted as an attempt to apply the principle of “phased regulation”: first, basic conditions and incentives, then fine-tuning as practical experience accumulates and the most pressing issues become clearer.

An important element of the draft law is the categorization of AI models: sovereign models, fully developed by Russian legal entities and operating on domestic infrastructure, and national models, which may include open-source components but are largely created by Russian developers. Both categories are required to comply not only with current legislation but also with “traditional spiritual and moral values.” From the standpoint of legal drafting, such wording is evaluative in nature and potentially creates room for varying interpretations, which may complicate law enforcement and lead to inconsistent oversight practices.

Particular significance is attached to the provision supporting developers, including by granting access to data necessary for training models. Access to high-quality and representative datasets largely determines the competitiveness of AI solutions. However, a conflict arises here: on the one hand, businesses (including small and medium-sized enterprises) need the broadest possible access to data; on the other hand, issues of confidentiality, personal data protection, and intellectual property rights must be taken into account. The Ministry of Digital Development has promised to prepare a separate document establishing a general approach to working with datasets, which indirectly confirms that the current version of the draft law deliberately avoids excessive detail so as not to hinder industry development with overly restrictive measures.

Thus, the current version of the draft law reflects a pragmatic approach: instead of attempting to immediately close all legal gaps, the state creates basic conditions for AI development and leaves room for regulatory fine-tuning. However, it is precisely this “fine-tuning” that will become the determining factor in the success of the entire initiative. If secondary legislation is adopted without sufficient consideration of the interests of all parties—businesses, developers, authors, and end users—there is a risk that the legal framework will either be too rigid and stifle innovation or too vague and fail to effectively protect the rights of market participants. Therefore, the coming months will be a critical period for shaping the actual legal regime for AI in Russia: the extent to which secondary regulations prove balanced and predictable will determine not only technological development but also public and business trust in AI.

The most controversial aspects—copyright and liability for AI-generated outputs—have been deliberately left outside the law itself and will be regulated later at the level of secondary legislation. This strategic choice can be viewed as a managerial decision: to allow the industry to develop without waiting for the full resolution of legal conflicts, while simultaneously preparing the regulatory framework based on real-world practice. Nevertheless, the absence of clear rules already creates risks for market participants: developers cannot confidently plan the use of third-party data and content, and users of AI services cannot fully assess the potential legal consequences of their actions.

The position of parliament reflects an understanding of this dilemma. Speaker of the State Duma Vyacheslav Volodin explicitly stated that deputies are effectively delegating to the government the authority to adopt secondary legislation, while retaining their oversight functions. As early as autumn, hearings are planned to review reports on the implementation of the law and to discuss draft secondary regulations with experts, business communities, and legal professionals. This indicates an intention to maintain parliamentary oversight of the process and to prevent excessive concentration of rulemaking power in the hands of the executive branch.

Special attention should be paid to the need to strike a balance between the unhindered development of domestic AI and the protection of authors’ interests. Indeed, the mass use of copyrighted works for training neural networks without the consent of rights holders may undermine incentives to create new content and, in the long term, deprive AI systems of high-quality training data. Finding a compromise here is one of the key tasks facing lawmakers.

Author: Candidate of Economic Sciences, Associate Professor, Department of World Economy and International Finance, Financial University under the Government of the Russian Federation Natalya Ivanovna Chovgan.