The EU Intellectual Property Office (EUIPO) has revealed more details about its Copyright Knowledge Centre, set to launch in November. Designed to become a “copyright home in Europe”, the centre comes at a time when the use of copyrighted works to train generative AI (Gen AI) systems is raising urgent legal and ethical concerns.
The centre, originally announced in the office’s Strategic Plan 2030, aims to equip copyright holders with clear, practical information on the use of copyright-protected content to train AI systems.
The details of the centre form part of a 436-page report published by the EUIPO on 12 May – ‘The development of Generative Artificial Intelligence from a copyright perspective’ – aimed at clarifying how Gen AI systems interact with copyright technically, legally and economically.
The report says the centre will help copyright holders manage and protect their intellectual assets. It also will provide a platform for discussing ways to support simple and effective mechanisms for copyright holders to reserve their rights, and for AI developers to comply with such reservations.
Drawing on the insights of this study, the centre will provide a foundation for discussions among experts on how copyright can effectively support content creation and innovation in the Gen AI landscape.
The study was prepared by a research team at the University of Turin Law School and the Nexa Center for Internet & Society from the Polytechnic of Turin.
It examines how copyright-protected content is used in training models, what the applicable EU legal framework is, how creators can reserve their rights through opt-out mechanisms, and what technologies exist to mark or identify AI-generated outputs. It also explores licensing opportunities and the potential emergence of a functioning market for AI training data.
Among some of its insights explored, the report notes that while most Gen AI developers source and use content available online without prior authorisation of copyright holders, a “market for direct licensing is slowly emerging”.
A number of rightsholders have entered into agreements with AI providers and even more are signalling their willingness to enter into negotiations. Stock image website Shutterstock, which claims to manage more than 530 million digital assets, has entered several licensing deals with major AI developers.
OpenAI has secured agreements with a growing number of major media companies, particularly in the press publications sector including Associated Press, The FT (Financial Times) Group and Axel Springer.
Article 4 of the EU’s Directive on Copyright in the Digital Single Market allows for reproduction and extraction of lawfully accessible works for text and data mining (TDM) purposes, unless explicitly restricted by rightsholders. This means that TDM can be used for commercial purposes, but rightsholders can opt-out of having their works used for TDM by clearly indicating their restriction.
According to the study, the structure of Article 4 creates the conditions for a possible market for licensing permissions for commercial TDM uses, contingent on rightsholders exercising their right to opt-out. AI developers need to “to engage more actively with the creative and cultural sectors on effective ways to account for the text and data mining opt-out expressed”.
The research took place between September 2024 and March 2025, and was supported by extensive desk research as well as interviews from key stakeholder groups including copyright holders, AI companies, technology solution providers and public organisations.
João Negrão, the EUIPO executive director, presented the report on 12 May at the European Parliament. He said the report “provides policymakers, rightsholders, AI developers and intermediaries with concrete data, legal analysis and recommendations on how IP offices such as the EUIPO can play their part in improving the status quo”.
He added: “It highlights where legal certainty is needed, where technical and licensing solutions can be improved and where deeper cross-sector dialogue is essential. This is not just a study – it is a call to coordinated, forward-looking action.”
The EU does not have a central copyright office in the way that some countries do, like the US with its Copyright Office. Instead, the EU has a system of harmonised copyright law across member states.
The EUIPO is already responsible for establishing and managing the EU orphan works database (since 2012) and the public single online portal for out-of-commerce works (since 2019).
US
The launch of the EUIPO study follows reports over the weekend that US President Donald Trump has fired the head of the US Copyright Office, Shira Perlmutter. Reuters reported that the top copyright official was fired over email on Saturday.
It came just two days after Trump fired the librarian of Congress, Carla Hayden. The Copyright Office is part of the Library of Congress.
Trump’s latest purge follows the US Copyright Office releasing the third and final installment of its report on copyright and artificial intelligence on 9 May.
The report outlines the use of copyrighted works in the development of Gen AI, concluding that copyrighted works are likely considered fair use when an AI model is deployed for analysis and research.
In what could be bad news for AI developers, however, the report also found that fair use likely does not protect the commercial use of copyrighted works to produce content that competes with its original works in an existing market.
On hearing of the latest dismissals, New York Congressman and Democrat Joe Morelle issued a statement saying that Trump’s termination of Register of Copyrights Shira Perlmutter “is a brazen, unprecedented power grab with no legal basis. It is surely no coincidence he acted less than a day after she refused to rubber-stamp Elon Musk’s efforts to mine troves of copyrighted works to train AI models”.
Email your news and story ideas to: [email protected]
