by Kaisa Ranta, based on a photo by Mister_fr CC-SA
Computer scientist Prof. Dr. Sebastian Stober (University of Magdeburg) and legal scholar Prof. Dr. Tim W. Dornis (University of Hannover) have worked in tandem to produce an expert opinion on the technological and legal aspects of training generative AI models, with a focus on the copyright implications of the processing of protected material for AI training.
In examing the technology of generative AI models in detail, the two professors have reached the conclusion that the training of such models is more than a simple case of text and data mining. It involves clear copyright infringement, for which no exception exists under German and European copyright law.
The research was funded by Initiative Urheberrecht (authors’ rights initiative) which represents the interests of approximately 140,000 German and European authors and performing artists in many fields including fiction and non-fiction, visual arts, drama, game development and many more.
Read more about the report here. (https://urheber NULL.info/diskurs/ai-training-is-copyright-infringement)