"There is a heated debate around this approach,

[tanjimajuha20]

Senior partner of O2 Consulting LLC Tatyana Safonova agrees with him. She noted that the issue of protecting the intellectual rights of content owners has not yet been fully worked out. And the fact that the owner and user of the content can be in different countries and different jurisdictions makes it even more complicated.

"I am far from thinking that by ticking the box and allowing the company to use its content for training AI, the owner actually enters into an agreement with it for the gratuitous transfer of non-exclusive license rights to intellectual property (by the way, in Russia you have to pay tax on gratuitously received brazil telegram income). Most likely, there is no such legal framework under the hood of YouTube. Thus, the ability to tick such a box gives copyright holders at least some, albeit ghostly, opportunity to express their will in relation to their intellectual property. Although it is unclear why this is done only in relation to the use of video content only for training AI, and not for any use in general," said Tatyana Safonova.

Nextons JSC lawyer Ekaterina Tytyuk explained that there is a simple rule in intellectual property law, according to which only the copyright holder or persons who have permission to do so can use the results of intellectual activity (RIA). This rule also applies to the use of RIA for training neural networks.

specifically in connection with the issue of AI training: AI developers cite the special importance of gaining access to data for the purposes of developing new technologies and periodically come forward with initiatives to develop exceptions to this rule. On the other side are the authors of RIAs, who believe that, despite existing prohibitions, large corporations are receiving unlimited access to data for training without appropriate permission and incorrectly interpret the provisions that allow the use of RIAs without consent in exceptional cases - for example, they cite the provisions on citation," she said.

According to Irina Zemkina, a leading lawyer at Enforce Law Company, there are no disputes in Russia yet related to the use of authors' works for training AI. She noted that in any case, the rights to the content are retained by the copyright holder, and if the latter wants to give consent to the use of the company's materials - the AI ​​developer, the relationship between them will resemble a license. Then the copyright holder will have the right to demand remuneration for the use of the work. The absence of such consent may raise doubts about the legality of use.

"On the part of the service, creating a mark about permission or prohibition of using videos by third parties for training AI is a step towards creating tools to protect authors from illegal use of videos for training models. Of course, it will not protect the video from copying, but will allow the author to have additional confirmation that he did not give third parties the right to use the content, if he needs such confirmation," said Oksana Fedorova, senior legal adviser of the ANO Institute of Artificial Intelligence (AIRI).