Ograničenja autorskog i srodnih prava: prepreka ili podsticaj razvoju veštačke inteligencije

In recent years, the AI-based services have increasingly become part of our everyday life. The AI systems are trained on third party’s copyrighted works and subject-matter of related rights, particularly on databases. This fact leads intellectual property lawyers to examine the legality of such use of third party’s protected works and subject-matter of related rights. The first question to be answered is whether the access, analysis and mining may infringe copyright and related rights. The second question logically follows and leads us to examine whether such uses may be permitted under copyright exceptions and limitations, prescribed by the law. The paper shall start with the analysis of the copyright exceptions and limitations prescribed in Serbian law, which may be of some, at least indirect, relevance to AI developers: (i) temporary acts of reproduction of copyrighted works; (ii) limitations related to the use of software; and (iii) quotations. The exceptions and limitations of a related right of database producer shall be analyzed separately, both from the perspective of Serbian law and European Union law. Finally, the author will assess the exception and limitation for the purpose of text and data mining, prescribed under the Directive on copyright in the digital single market, adopted in 2019. The analysis will particularly focus on the manner and moment in which the copyright and related right owners may prohibit commercial text and data mining.