GESAC’s statement in view of the upcoming Trilogue meeting

In view of the upcoming Trilogue meeting on October 24, GESAC has issued a statement, urging European legislators to prioritize the inclusion of record-keeping and transparency obligations regarding the use of copyright-protected content by generative AI models. These obligations are necessary to ensure responsible and trustworthy AI systems, level the playing field, and enable European innovation and artistic creation.

Read on to find out more.

Ai Chip Artificial Intelligence, Future Technology Innovation

The introduction of record keeping and transparency obligations on the use of copyright protected content by generative AI models is necessary to ensure responsible and trustworthy AI systems, level the playing field, and enable European innovation and artistic creation.

Brussels, 20 October 2023 – GESAC, which groups 32 European authors’ societies (CMOs) and represents as such more than one million authors from the various artistic fields, music, audiovisual, visual arts etc., welcomes the recognition of the need for dedicated record keeping and transparency obligations to support enforcement of Union copyright law in the Spanish Presidency’s recent document that proposes, among other important elements, possible ways to introduce new rules for foundation models and general purpose AI systems.

Human creation – music, films, books, visual arts etc. – are essential elements that fuel AI based services and ensuring the quality of their output for the benefit of the public. Authors’ societies and the creative sector are prepared and willing to embrace these technologies and their development through licensing and cooperation. However, without the necessary record keeping and transparency obligations in the AI Act, implementing the current EU and national laws that provide creators with economic, moral and personality rights would simply be impossible. The Spanish Presidency’s document rightly underlines that generative AI models should “demonstrate that they have taken adequate measures to ensure the models are trained in compliance with the applicable Union copyright law, in particular the opt-out from the TDM exception” and provides useful considerations on the governance and enforcement of such possible obligations. It is also key, as referred to in the Spanish Presidency’s text to provide that the output is detectable as artificially generated or manipulated.

Transparency and record keeping are the basic standards for any business that operate in the digital environment and the record keeping of the input is already undertaken by the relevant services/systems for their own purposes. It is, therefore, only common-sense to expect from all services, regardless of where they are based, or where the models they use are developed and trained, to provide the relevant information that would allow identification of protected works to the requesting rightsholders.

We call on the Member States, the European Parliament and the European Commission to agree on the inclusion of this critical issue in the final text and we look forward to cooperating with the co-legislators for the consideration of an appropriate and effective wording on this issue as an immediate next step.

Please contact for further information:
Véronique Desbrosses – General Manager