Paul George Buta**

Abstract
The issue of copyright infringement by artificial intelligence (AI) has become more salient, with the worldwide popularity of “Heart on My Sleeve” and with two other class action claims having been filed. The current piece aims to set out the core concepts, to list the possible issues to be addressed by further research and to suggest some perspectives to so address this matter. The most relevant from a copyright infringement perspective seem to be: (1) the distinction between inputs and outputs, i.e. the data the model is trained on and the product of the model; and (2) the existence of different models, each with a different level of human involvement and the existence, within each such model, of component algorithms, each with a different level of human input and relative importance of such. Taking all this into account clarifies the issues that could be incidental to an analysis of possible risks of copyright infringement by artificial intelligence models and also provides some guidance as to the perspectives from which such analysis could be conducted.

** Conferențiar universitar doctor, Facultatea de Drept, Universitatea “Nicolae Titulescu” din București (e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.).