Copyright Protection for Output Resulting from Use of Generative AI Systems – Some Reflections
By Ole-Andreas Rognstad[*]Professor dr. Juris, Department of Private Law, Faculty of Law, University of Oslo.
1. Introduction
Bernt Hugenholtz has in this volume given an excellent overview of recent case law and the discussion of “the state of the art” regarding copyright protection of output resulting from the use of generative AI (GAI) systems, including the copyright status of “prompts” at the boundaries of ideas and expression.[1]P. Bernt Hugenholtz, “Prompts Between Ideas and Expression. Generative AI and Works of Authorship”, NIR 2025, 125–132. See also on the same topic the more extensive analysis in P. Bernt Hugenholtz, “Copyright and the Expression Engine. Ideas and Expressions in AI-assisted Creations”, Chicago Kent Law Review 2024 (forthcoming), available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4982516 tz :: SSRN. His intervention demonstrates the different approaches and attitudes to the question, and not least the diverging outcomes in two leading AI regions – the US and China. It also ends on a note regarding the pending cases before the CJEU on the originality and similarity requirements in EU copyright law, and especially the questions raised by the Swedish Appellate Court in case C-580/23 Mio, expecting that the answers to these questions “are likely to inform the future protection of AI creations as well”.[2]Hugenholtz 2025, 132.
In this article I will not reiterate the findings and the positions of the various courts and the discussion about the relevance of the idea/expression dichotomy to prompts but refer to Prof. Hugenholtz’ presentation about this. Nor will I attempt to predict the outcomes of the pending CJEU cases. Rather, I will present an approach to the problem that I believe is in line with the current legal situation in the EU based on an article about copyright protection of AI generated content written before the emergence, or at least the general use, of GAI.[3]Ole-Andreas Rognstad, “Creations caused by humans (or robots)? Artificial Intelligence and causation requirements for copyright protection in EU law”, in Taina Pihlajarinne and Anette Alén-Savikko, Artificial Intelligence and the Media. Reconsidering Rights and Responsibilities, Edward Elgar 2022, 172. Slightly in contrast to the findings of the 2020 “Hugenholtz report”, which concluded that the “existing EU framework of copyright and neighboring rights can adequately address the main problems raised by AI-assisted production”,[4]As phrased by Hugenholtz 2024, 2. See P. Bernt Hugenholtz et al, “Trends and Developments in Artificial Intelligence: Challenges to the Intellectual Property Rights Framework”, study prepared by IViR and JIIP, Final Report for the European Commission 2020, available at https://op.europa.eu/en/publication-detail/-/publication/394345a1-2ecf-11eb-b27b01aa75ed71a1/language-en. See also P. Bernt Hugenholtz and Joao-Pedro Quintais, “Copyright and Artificial Creation: Does EU Copyright Law Protect AI-Assisted Output?”, International Review of Intellectual Property and Competition Law (IIC) 2021, 1190–1216, and Johan Axhamn, “Copyright and Artificial Intelligence – with a focus on the area of music”, in Festskrift til Jørgen Blomqvist (eds. Rosenmeier et al.), Ex Tuto Publishing 2021, 33–86. my conclusion was that the framework does apply, but that its application creates considerable problems not least in regard to determining the scope of protection of AI generated output.[5]Rognstad 2022, 189–191. This problem is rather accentuated than diminished after the emergence of GAI.
