The government is discussing new regulation of work with neural networks

The Ministry of Industry and Trade, the Ministry of Digital Development and the Alliance in the Field of Artificial Intelligence must prepare proposals for the Federation Council within a month to finalize legislation regarding the allocation of industrial data into a separate category, the rules for their use for training AI and determining the rights to final decisions. Corresponding amendments to the regulation of copyright, in this case for products created by neural networks, should be made to the Civil Code of the Russian Federation in 2024. But releasing industry data to train neural networks risks violating trade secrets, experts warn.

The Federation Council recommended the Ministry of Industry and Trade, the Ministry of Digital Development and the Alliance in the field of artificial intelligence (which unites Sberbank, Gazprom Neft, Yandex, MTS, VK, etc.) to work on a number of initiatives aimed at developing AI in industry, the deputy chairman told Kommersant Council for the Development of the Digital Economy under the Federation Council Artem Sheikin.

Thus, the ministries and the alliance should prepare amendments to the fourth part of the Civil Code aimed at regulating the circulation of the results of intellectual activity created using AI, including in industry, since there is currently “no legal certainty” in this direction.

“We are talking about AI technologies that process information and provide results – for example, text, calculations, analysis in the field of industrial data,” explained Mr. Sheikin.

But to implement the project, it is first necessary to establish what data should be considered industrial so that it can be used in AI training. To do this, the Federation Council recommended that the Ministry of Industry and Trade, the Ministry of Digital Development and the alliance work out criteria for classifying data as industrial, the senator explained: “This could be data from technological processes, results of remote sensing of the Earth, information about minerals or data in the field of metallurgy and petrochemistry.”

The recommendations were given in pursuance of the president’s instructions dated July 13, within the framework of which the national project “Data Economy” until 2030 must be approved by July 1, 2024, the Federation Council noted. They clarified that in November they held a round table on this issue with the participation of authorities and experts, including Gazprom Neft: “Recommendations will be sent in the coming days, the response period is a month.” The Ministry of Digital Development confirmed that they took part in the round table. The Ministry of Industry and Trade did not respond to the request; the AI ​​Alliance and Gazprom Neft declined to comment.

Mikhail Fokanov, a leading expert on AI tasks at Tsifra Group, explains that at the moment, AI-based technologies are still tools that make it easier to create products. Therefore, Mr. Fokanov believes, it is advisable to establish ownership of the product created by the neural network, and assign this right to the one who used it to create the product, and, for example, not to the one who created the tool itself.

However, the leader of the Fundamental Services cluster of the Sfera platform (T1 holding) Valery Nikanorov believes that “the initiative will definitely raise some doubts among companies.” “Data on mineral resources, for example, may be commercial developments that businesses will not want to share for use in training AI models,” Mr. Nikanorov believes.

If the information makes it possible to create an AI product that provides a competitive advantage to the owners of primary data, then transferring it to third parties may bring losses to the company, Tsifra Group agrees. Without further processing, Mr. Fokanov explains, such products “will have the characteristics of a trade secret and, by default, are protected by sections of the law regarding the protection of intellectual property.”

Tatiana Isakova