From August 2025, the EU AI Act[1] will require providers of generative AI (Gen AI) systems (Providers) to be much more transparent about the content they have been harvesting and ingesting to train these systems. This requirement will de facto apply to all Providers worldwide, not just those within the EU.

A significant report[2] published in France in December 2024 (Report) sets out the likely form and content of a template that Providers will have to complete. This template will give rightsholders, including those holding copyright (such as newspapers, publishing houses and film studios), neighbouring rights (such as musicians and actors) and personality rights (such as sportspersons, celebrities and influencers), information that will help them file infringement claims in court against those same Providers. Although the Report was addressed to the French Ministry of Culture, it is expected to influence discussions with the European Commission, and therefore have impact at EU level. 

It is widely known that the film and television, sports and advertising sectors are increasingly using and relying on  Gen AI systems. Increasingly, businesses operating in such sectors are also finding that their own content has been and continues to be used in Gen AI systems. As such, the Report, which is focused on transparency, is likely to herald a new era of infringement claims in all directions. 

Many commentators have noted that prominent rightsholders, such as the Hollywood studios, have adopted a 'wait-and-see' approach in relation to this issue. These organisations may now start to feel that the time is right to argue that their valuable IP has been used by Providers without authorisation. 

The new and detailed requirement for transparency set out in the Report could be compared to the discovery phase of proceedings, allowing rightsholders to formulate a strategy regarding how they threaten or commence litigation and engage in negotiations. It may also result in them opting out entirely from the Gen AI training process.

Providers will likely argue that the way they break up and filter content will affect the merit of any infringement claims brought, and/or that the text and data mining exceptions in Directive EU 2019/790[3] already provide legal cover for their activity. However, the Report zooms out to the larger EU policy level, suggesting an inevitable market-based and less contentious approach, similar to that applied when collecting societies administer royalty payments for the music, film and television industries. 

A further report is expected in early 2025 looking more closely at these related payment and monetisation issues. It is expected to advocate for this latter approach as best serving the EU market and innovation, even though this is closer to the more laissez-faire Japanese and US approaches to Gen AI system training than might be expected from an EU IP law perspective.

The scope of the Report does not extend to the usage of Gen AI systems by content creators. 

We will share further updates on this matter when we have them. In the meantime, if you would like any further advice or assistance, please do get in touch.
 

[1] Regulation (EU) 2024/1689 of the European Parliament and of the Council on harmonised rules on artificial intelligence.

[2] Rapport de mission relative a la mise en œuvre du reglement europeen etablissant des règles harmonisees sur l’intelligence artificielle, Alexandre Bensamoun, Lionel Ferreira, Frederic Pascal (9 December 2024).

[3] Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market and amending Directives 96/9/EC and 2001/29/EC.