EU consults on copyright landscape in the age of generative AI

The European Commission has launched a new “targeted” consultation on EU copyright laws, which is intended to support its review of the 2019 Directive on copyright in the Digital Single Market (DSM Directive) and to inform potential changes to the EU’s copyright framework, including in the context of generative AI (GenAI).

Background

This Consultation comes off the back of the European Parliament’s resolution on copyright and GenAI, which was adopted in March, and the Commission’s subsequent call for evidence (CFE) that was published in May (see our summaries in the July edition of The IP Brief and our blogs here and here).

The Consultation builds on the responses to the CFE, seeking further feedback on the main issues that rightsholders and GenAI providers face in this context, as well as on potential mechanisms to address them (many of which reflect those put forward by the European Parliament in its resolution).

That said, the Commission has made it clear that the mechanisms set out in the Consultation are not an exhaustive list of the options under consideration, nor are they to be read as an indication of the Commission’s preferred course of action. Rather, the options put forward are intended to reflect what is being discussed more broadly in the public discourse and among EU member states and lawmakers. 

The Consultation is open for submissions until 3 November and will inform the Commission’s preparation of a targeted initiative for potential future reform. 

Summary of the Consultation

1. GenAI and copyright

As readers of this blog will be aware, the DSM Directive includes two exceptions for text and data mining (TDM) – a broad one that permits the use of copyright works absent a rightsholder opt-out (Broad TDM Exception) and a narrower one for scientific research. The Consultation acknowledges the challenges that rightsholders who have opted their rights out of the Broad TDM Exception face in enforcing their rights, licensing out their works and being paid for the use of their content, whilst at the same time recognising GenAI providers’ need for legal certainty and access to high-quality training data.

The Consultation also highlights relevant AI Act obligations on general purpose AI (GPAI) model providers to put in place a policy to comply with EU copyright law and publish a training data summary using the AI Office’s template (see our briefing here) and notes the challenges that remain in ensuring effective implementation of those provisions.

The Commission is therefore exploring ways to address those challenges and seeking to gather stakeholders’ views on (i) the issues facing different stakeholders in the context of GenAI and copyright; (ii) the barriers faced by both rightsholders and GenAI providers; and (iii) possible mechanisms that could be introduced to address these issues, and, in each case, the impact (or potential impact) on the respondent. 

Of most interest are the proposed mechanisms that are put forward, noting that most have been mentioned before. These include:

• On the copyright enforcement side: (i) standardised TDM opt-out protocols and a centralised opt-out registry; (ii) clarification of EU copyright laws’ application to the use of copyright-protected content for AI training and other uses; (iii) web crawler disclosure requirements; (iv) additional obligations on GenAI providers (e.g., in relation to record keeping, independent audits and use of protected content); (v) court disclosure obligations; and (vi) a rebuttable presumption of use of content where existing transparency obligations are not fulfilled. 
• On the licensing side: (i) measures to support a licensing marketplace; (ii) measures to aid licence negotiation (e.g., model clauses, structured mediation and/or promoting collective licensing); and (iii) introducing fair compensation obligations or remuneration rights in favour of rightsholders.

2. Digital likeness and deepfakes 

The Consultation also includes a section on AI-generated imitations of performers (also mentioned in the CFE), which looks to examine whether current protections adequately enable performers to control and exploit their digital likeness, and seek views on (i) potential additional measures and (ii) whether copyright law is, indeed, the right place to address this issue.

This issue has gained prominence recently, with high-profile performers and sports stars (such as Taylor Swift, Cole Palmer and Luke Littler) filing trade marks to protect their voice and likeness from AI misuse. In the UK, a government consultation on digital replicas is also expected (see blog), noting that this had been slated to land over the summer.

3. Other aspects of the Consultation 

It is worth noting that the Consultation is not limited to the challenges posed by copyright and GenAI. As envisaged by the CFE, it also covers a number of other copyright areas, including: online piracy of time-sensitive content, including live events; remuneration rights for non-EU music producers and performers; and exceptions for scientific research.

Comment

Although the Consultation and its timing are tied to the DSM Directive (which required a review “no sooner than 7 June 2026”), the decision to consult now - while the AI Act is still so new - is notable. It’s also interesting to see the Commission already starting to float potential changes to some of the obligations that have been imposed on GPAI model providers under the AI Act. For example, one of the proposed mechanisms to support enforcement of copyright is to oblige GPAI model providers to confidentially disclose information to rightsholders on request “as an alternative” to the need to publish the top 10% of crawled website domains under the AI Office’s template for summarising training data (see blog) - perhaps suggesting the Commission is assessing early AI Act compliance practices to shape its copyright approach. 

The timing of the Consultation is also significant given pending EU copyright cases involving AI, including Like Company v Google and the recently referred questions in Manchester United Fanatics Klub v Google, as well as a further expected referral from the German courts in the Robert Kneschke v LAION case. We’re also still waiting for the outcome of the consultation on TDM opt-out protocols that the Commission launched in late 2025 in the context of the AI Act.

In the UK, the government has opted for a ‘wait-and-see’ approach following a change of tack earlier this year (blog here). By contrast, this Consultation appears to be more actively preparing for possible legislative measures to address the challenges that the Commission acknowledges exist under the current framework. Although we will have to ‘wait-and-see’ where the Consultation lands, meaningful reforms may emerge from the EU sooner than the UK.