Journal of
Italian
Banking
Association

year 100
 

Archive » July / August 2024 » Artificial Intelligence and Copyright. Training an AI model and the means to avoid Data scraping

Artificial Intelligence and Copyright. Training an AI model and the means to avoid Data scraping

Cecilia Trevisi
July / August 2024 - n. 7-8
Keywords: Intelligenza artificiale, diritto d'autore, data scraping
Jel codes: O33, K10

In the training phase of an Ai model, copyright-protected content organised in structured datasets is used, which may contain pre-existing works that are then, in the case of generative models, recomposed into a hybrid, elaborated form. The developer of such models and Ai trainers should seek the consent of, and pay a fee to, all right holders of the materials used to make such use. Nevertheless, during the development and training of Ai models, permission for such uses may not be necessary in the case of text and data mining carried out for scientific research purposes, and for other purposes provided that, in the latter case, the opt-out has been properly exercised.

Interested in this paper?
Buy the issue