EUIPO Study on Generative AI and Copyright – Key Takeaways and its relevancy for Switzerland

Therefore, the European Union Intellectual Property Office (EUIPO) has just released a landmark study on the impact of GenAI on copyright law. The report offers not only an in-depth look at how GenAI is challenging the foundations of copyright law across Europe but a detailed analysis of the technical, legal, and economic challenges that GenAI poses, particularly in relation to the use of copyrighted content in AI training and generation. With GenAI services rapidly transforming the creative and technology sectors, the study is set to guide future EU regulation and influence both public and private sector strategies.

The study emphasizes that copyright remains central to the EU’s creative economy, accounting for nearly 7% of GDP and more than 17 million jobs. However, while innovation is encouraged, GenAI’s reliance on vast quantities of web-scraped data – including copyright-protected material – necessitates a re-evaluation of legal frameworks, especially concerning the rights of creators and copyright holders.

Main Findings from the EUIPO Study

1. The EU Copyright Directive as well as the newly enacted EU AI Act allows rights holders to opt out of the text and data mining (TDM) used for AI training and thereby practically urging GenAI developers to seek licensing agreements (cf. point 3 below). However, there is currently no harmonized or enforceable mechanism to express these reservations. Existing technical and legal tools – such as the Robots Exclusion Protocol or contract-based terms – are fragmented, non-standardized, and often difficult to implement. This creates uncertainty for both creators and AI developers. While no single standard prevails, a combination of legal and technical measures is currently the best practice.
2. Despite new obligations under the EU AI Act for content transparency, there is no universal standard for labelling or detecting AI-generated content. This complicates copyright enforcement, public trust, and content verification, and highlights the need for interoperable technical solutions such as provenance tracking and machine-readable disclosures.
3. While lawsuits regarding AI training data are increasing worldwide – including in Europe – there is a parallel trend toward private licensing agreements between rights holders and GenAI developers. These markets show promise, particularly for sectors such as journalism and scientific publishing. Smaller rights holders may also benefit from aggregation services that facilitate content licensing at scale. However, they may risk exclusion if licensing models are not transparent or scalable. For the moment, the development of a robust licensing market hinges on effective, enforceable rights reservation mechanisms.
4. The EUIPO advocates for the creation of federated databases to register TDM opt-outs and will launch a Copyright Knowledge Centre by the end of 2025. This centre will serve as an information hub and coordination platform for creators, AI developers, and policymakers to access practical tools, engage in licensing negotiations, and promote transparency around GenAI-generated content.

And the Swiss?

Although Switzerland is not bound by EU law, it is deeply interconnected with the European digital economy. The findings of this study are therefore highly relevant for Swiss businesses, especially those operating across borders or engaged in content creation, AI development, or publishing. Here’s why:

• Cross-Border Compliance and Legal Risk: Swiss content and data are routinely scraped and used to train GenAI models that operate in or target EU markets. Swiss companies may find themselves indirectly subject to EU legal frameworks or facing enforcement challenges abroad.
• Business Opportunities: Swiss media houses, publishers, and creative industries can strategically opt out of data mining and benefit from new revenue opportunities through licensing their content for GenAI training, particularly if they proactively manage their rights and engage with content aggregators or licensing platforms, thereby opening new revenue streams and protecting intellectual assets.
• Innovation and Interoperability: For Swiss tech developers, compliance with EU standards on transparency, data usage, and content provenance may soon become a de facto requirement for market access and partnership with EU-based entities.

Regulatory Influence: As Switzerland crafts its own digital and AI strategies (current status: https://www.bj.admin.ch/bj/de/home/staat/gesetzgebung/kuenstliche-intelligenz.html), the EU model offers valuable guidance. Participation in international regulatory discussions -especially via WIPO and bilateral treaties – will be critical to ensuring Swiss interests are protected.

Conclusion

The EUIPO’s study (available here: Study – The development of GenAI from a copyright perspective – FULL report.docx) is a timely and comprehensive overview of the legal and regulatory challenges – and opportunities – posed by GenAI from a copyright perspective.

MLL Legal will continue to monitor the evolution of EU and Swiss law in this area and is available to provide tailored advice on how your organisation can navigate these emerging legal complexities.

For tailored advice on GenAI and copyright, please contact the MLL Legal team.