The EU copyright consultation has been extended to 10 November, and the compliance boundary for generative AI training data may be redrawn. The European Commission launched the targeted consultation on 29 September; it was originally due to close on 3 November and has now been extended to 10 November. The consultation creates no new obligations directly, but it is addressed to rightsholders, generative AI providers, research organisations and other stakeholders, and the responses collected will decide whether the EU copyright framework needs to be strengthened. For companies that provide or train general-purpose AI models in the EU, this is a window to influence the direction of the rules, rather than waiting for legislation to arrive and reacting afterwards.
Four topics, and AI training is only one of them
The consultation covers four topics: the use of copyright-protected content in AI; action against online piracy of content such as live events; the application of the single equitable remuneration right for music performers and producers; and copyright in scientific research. The official notice says the consultation builds on a call for evidence earlier this year and on previous studies. The consultation document also lists more specific issues separately, such as generative AI imitations of performers and the use in the EU of sound recordings from non-EU countries, showing that the discussion has reached the detail of licensing and enforcement. Companies cannot focus only on training data; rule changes in music, live events and research could also affect product design.
The current rules have not changed, but the obligations are already running
A consultation is not a legislative amendment, and the current EU copyright framework remains in force. Under the EU AI Act, providers of general-purpose AI models must adopt a policy to comply with EU copyright law and publish a sufficiently detailed summary of the content used for training. The text and data mining exception in the 2019 Copyright Directive allows mining where the rightsholder has not opted out in a machine-readable way. Companies must therefore meet their existing obligations today: keep records of training-data sources, respect opt-out declarations, and prepare a summary that can be published. Reading the consultation as a pause in the rules is the easiest mistake to make.
Three things companies should do before 10 November
First, check your own role: whether you are a model provider, a content rightsholder or a research organisation, since each faces different questions. Second, put the evidence in order: training-data sources, licensing agreements, records of how opt-outs were handled, and the basis for the summary; fill any gap first. Third, decide whether to submit a response: companies providing services in the EU are equally affected by future rules, and a submission should be built around specific facts and workable proposals. The deadline is 10 November, and responses submitted late may not be taken into account.
The risk sits at the boundary: this is an EU rule, not a global one
This consultation applies in the EU and cannot simply be applied to the United States or China. A US court of appeals has already rejected a fair-use defence in Thomson Reuters v. Ross, and the US court's approach follows a different path from the EU text and data mining exception. A company that treats a US judgment as an EU compliance conclusion, or treats EU consultation views as a global standard, will misjudge the risk. After the consultation, the European Commission also plans to present an AI strategy for the cultural and creative sectors in the first quarter of 2027, so the direction of the rules may still change; companies should comply with the obligations already in force and then track what follows.
