skip to Main Content

AI-Generated Database

AI-Generated Database

Do the creation/obtaining dichotomy and the substantial investment
requirement exclude the sui generis right provided for under the EU
Database Directive?
Reflections and proposals*

Paolo Burdese



The starting point of this analysis is the position of the European Commission, which, in its 2018 evaluation of the Directive, clearly stated that the sui generis right does not apply broadly to the data economy, including Artificial Intelligence (AI). This position was justified by the spin-off theory and the lack of substantiality of the investment involved in creating the databases concerned. According to Article 7 of the Database Directive, in order to benefit from sui generis right protection, the maker of a database has to make a qualitatively or quantitatively substantial investment in either obtaining, verifying or presenting the contents of the database. While the CJEU and domestic courts agreed that the threshold for the substantiality of the investment should be low, major issues arose in connection with the aim of this investment. Indeed, due to judgments handed down by the CJEU in 2004, investment in creating the content of a database cannot be interpreted as “obtaining” such materials and, therefore, is not relevant.

This interpretation created the so-called creation/obtaining dichotomy. However, while these judgments significantly reduced the number of protectable databases, the CJEU clearly rejected the spin-off theory. Several years later, the CJEU’s judgment in the Ryanair case made the framework even more complex, the court ruling that the legitimate user of a database not protected under Article 7 of the Directive cannot benefit from the rights granted under Articles 8 and 15. This decision gave rise to a true paradox, because an unprotected database can benefit from stronger
protection by contractual arrangement. Looking beyond the legal framework, this paper challenges the view of the European Commission and asserts that, as a matter of principle, AIgenerated databases can be protected under the sui generis right. Furthermore, the fact that this kind of database is usually generated by datarecording or data-mining processes, which involves obtaining rather than creating, means that the creation/obtaining dichotomy is no longer tenable and, therefore, should be abandoned. Moreover, rejecting this dichotomy would also mitigate the negative outcomes of the Ryanair decision.
However, it seems clear that granting easier protection in this way might generate serious access-related issues, especially where sole source databases are concerned. What is needed is a balance that, echoing the 1992 Proposal of the Directive, could be achieved by introducing compulsory licensing provisions.



***Clicca qui per continuare a leggere il contributo***

Back To Top