While organisations representing rightsholders in Brussels are vastly outnumbered by large tech companies, EU Member States continue prioritising licensing mechanisms and the role of collective management organisations (CMOs). Following the Hungarian and Polish EU Presidencies initiative to advance the discussions among member states on AI and copyright (read part 1), Denmark followed up with a questionnaire about licensing frameworks. Here’s what we’ve learned (and responded):
So far, legal proceedings have been initiated in three EU countries.
In France, the National Publishing Union, the National Union of Authors and Composers and the Société des gens de lettres sued Meta (Mar 2025) for training its AI without consent. In Germany, GEMA launched lawsuits against OpenAI (Nov 2024) and Suno AI (Jan 2025), claiming that authors songs are being used without providing remuneration. In Hungary, the Budapest court has referred a case to the CJEU (C-250/25, Apr 2025), concerning Google Gemini’s use of copyrighted works.
To date, only Spain and France have launched licensing initiatives.
In Spain, a proposal was presented (Nov 2024) to introduce extended collective licensing for AI uses. However, this was withdrawn due to opposition, and a promise was made that the government would facilitate a conversation between parties. In France, the CSPLA (the Superior Council of Literary and Artistic Property) is commissioning studies to inform discussions. The French Ministries of Culture and of Economy have launched a consultation process to bring rightsholders and AI companies together to share best practices for negotiating agreements and finding appropriate remuneration models.
Commercial licensing agreements have yet to be established for audiovisual authors in Europe.
The only licensing agreement between audiovisual operators and AI companies, of which the SAA is aware of is one concluded in 2024 by our member SACD in France with Genario AI, specialised in assisting with scriptwriting for audiovisual works. This agreement guarantees remuneration to audiovisual authors for the use of their works.
Recent studies by EU institutions, agencies and academia confirm the lack of such agreements in the audiovisual sector, continuing a trend where authors are not remunerated while AI companies grow their fortunes. So, what can be done? This is what the SAA replied to the questionnaire from the Danish Presidency:
EU and national pressure for AI companies to be transparent about their training data.
In short, we are disappointed by the AI Act implementation package presented in July (Code of practice, Template and Guidelines), as are other rightsholder organisations (see our joint reactions). It does not provide enough transparency, nor does it encourage licensing negotiations between rightsholders and AI companies.
Without transparency it is impossible to know which works have been used to train AI systems. AI companies claim that disclosing their training data would reveal trade secrets. Meanwhile, the opt-out principle remains vague and unworkable, effectively giving AI companies free rein to scrape any content they want, often bypassing paywalls or using pirated sources, without consequences. The robots.txt protocol and opt-out principle are inadequate: they place the entire burden on rightsholders to protect their works whilst offering no real safeguards against AI training.
To improve the situation, AI companies should at least disclose the works used to the rightsholders. Even if the training data were considered trade secrets, non-disclosure agreements, already standard practice between CMOs and users, could address this concern. The Danish expert group’s report on copyright and AI published in September 2025 supports this approach, recommending that rightsholders should have direct access to training data to identify the repertoire used and enter into licensing agreements.
Remuneration mechanisms for authors through CMOs, irrespective of the licensing model.
Regardless of the licensing model chosen, it is important to create a mechanism whereby authors receive the remuneration they are owed. CMOs are well-placed to ensure that remuneration reaches authors, even smaller ones, as they are used to managing the mass use of works and extensive repertoires. It’s their raison d'être. Therefore, as with any other use, it is the role of CMOs to manage AI.
It could be ECL, voluntary or mandatory collective management, an obligation to conclude a licensing agreement, or a statutory remuneration right. The most important thing is that the authors are remunerated. This can only happen if the legal framework clearly ensures authors’ remuneration with the support of CMOs.
Presumption that AI uses copyrighted content.
A presumption of copyright use would encourage the dialogue between AI companies and rightsholders, promoting transparency about the content used. AI companies would be required to license content unless they can prove that they have not used it. This could lead to more agreements between the parties, encouraging AI companies to disclose some of their training data in a way that protects their trade secrets while providing rightsholders with precise information about which of their works have been used. However, if uses continue to be presumed covered by the TDM exception, the impact would be limited.
Even when contacted by CMOs representing audiovisual authors, AI companies either do not reply or claim that they are protected by the TDM exception. These companies are not open to dialogue, and the lack of licensing, as recently reported by an EUIPO study, merely proves a situation that rightsholders have been highlighting for months, if not years now.
We therefore recommend that the EU Commission organises a high-level expert group to facilitate dialogue between AI companies and rightsholders. This is in line with the recommendations of Prof Lucchi’s study to the EU Parliament's Committee on Legal Affairs and the Danish expert group on copyright and AI.
Meanwhile, the MEPs of the Legal Affairs Committee are discussing MEP Voss’s draft report on copyright and AI, with a committee vote in December and a plenary vote expected for the beginning of 2026. The EU Parliament has long been the strongest ally of artists, and now more than ever, we need MEPs to support our call for a structured dialogue that ensures both innovation and fair remuneration for Europe's audiovisual authors (read our joint statement with FERA and FSE).
Read the full contribution below.
Public Affairs and Communication Director, SAA