Ground-breaking German study reveals AI Training is Copyright Infringement

---

Initiative Urheberrecht (Copyright Initiative), a German grouping of representative organisations across the cultural and creative sectors, commissioned Prof. Dr. Tim W. Dornis (University of Hannover) and Prof. Dr. Sebastian Stober (University of Magdeburg) to undergo an interdisciplinary study on the technological and legal aspects of training generative AI models, that was presented in the European Parliament on 5 September. The study finds that the training of Generative AI models does not fall under the text and data mining exception.

The report highlights that during generative AI training process, several acts of reproduction and copying of copyrighted works take place which are likely to be infringing on intellectual property rights. This includes not just data collection but also the internal processes within AI models, raising concerns about how generative AI Large Language Models (LLM) like ChatGPT handle copyrighted content.

The study outlines that the exception for commercial text and data mining is inapplicable: the training of generative AI models does not limit the use of the training data to a simple analysis of the “semantic” information contained in the works. It also extracts the “syntactic” information in the works, including the elements of copyright protected expression.

Although AI trained outside Europe might bypass local legislation, the deployment of such models for creative use in Europe triggers jurisdiction under EU and German copyright law.

The study also raises concerns about the future of human creativity, underlying that AI-generated content could overshadow human-produced work in cultural and creative industries. The study authors urge European lawmakers to balance the need for AI innovation with strong copyright protections, ensuring fair remuneration for creators.

 

To go further:

Dornis, Tim W. and Stober, Sebastian, Copyright and training of generative AI models – technological and legal foundations (in German): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4946214 

Press release, abstract & executive summary:
https://urheber.info/diskurs/ai-training-is-copyright-infringement