German collecting society GEMA now seems to be going on the offensive against providers of generative AI systems. Following the presentation of a – in their opinion – fair licensing model for generative artificial intelligence at the end of September, an “AI Charter” as a suggestion and guideline for the responsible use of generative AI was presented at the beginning of November, and now a lawsuit has been filed against OpenAI at the Munich Regional Court.

Subscribe to our IP Newsletter

We summarise what is already known about the case and the legal issues at stake.

1. GEMA's lawsuit against OpenAI

The Gesellschaft für musikalische Aufführungs- und mechanische Vervielfältigungsrechte (GEMA) is a collecting society for authors of musical works. It currently has over 95,000 members in Germany and represents the rights of over two million rights holders worldwide.

In a press release dated 13 November 2024, GEMA announced that it had filed a lawsuit against OpenAI with the Munich I Regional Court. The lawsuit is directed against both the US parent company OpenAI, L.L.C. and OpenAI Ireland Ltd. which operates ChatGPT in Europe.

The lawsuit is based on the allegation that OpenAI has trained its well-known product ChatGPT with copyrighted song lyrics, which are also taken from the repertoire of GEMA members. This was easy to prove, as ChatGPT could reproduce the lyrics in full without further access to the internet. In addition to the reproduction of the original lyrics, unauthorised adaptations (hallucinations) and infringements of moral rights were also detected. OpenAI had therefore systematically used copyrighted material to train its AI model, consciously accepting copyright infringements. However, the use of song lyrics was only permitted in return for appropriate remuneration, for which GEMA has developed what it considers to be a fair licensing model.

As in the proceedings brought by Robert Kneschke against LAION e.V. before the Hamburg Regional Court, the legal issue will essentially be the application of the “text and data mining exception” (“TDM exception”) in Section 44b of the German Copyright Act. GEMA is of the opinion that the requirements of this provision are not met: The TDM exception was already inapplicable and, moreover, GEMA had declared an effective reservation of use on behalf of its members. The use of song lyrics to train generative AI was therefore illegal in any case.

GEMA sees the lawsuit as a landmark case to clarify a number of open legal questions. In particular, it wants to refute the AI providers’ objection that the training of their AI systems is possible without consent and remuneration. The licence model developed by GEMA has not yet become established on the market, not least because the AI providers cite unresolved legal questions.

The case will raise a number of interesting legal issues. The Munich Regional Court will have to deal with the question of whether the TDM exception is applicable to the training of generative AI. In its LAION decision, the Hamburg Regional Court answered in the affirmative. If the Munich Regional Court follows this opinion, the interesting question will arise as to whether GEMA has declared an effective, in particular “machine-readable” reservation of use for its members pursuant to Section 44b (3) of the German Copyright Act – in this respect, nothing concrete has emerged from GEMA’s statements so far. In its (appealed) LAION decision, the Hamburg Regional Court recently indicated – rather surprisingly – that a reservation of use in natural language could in principle satisfy the requirements of Section 44b (3) of the German Copyright Act.

Against this background, this case is likely to attract a great deal of attention. Indeed, it may well become the landmark case on the issues it raises, not least because it is more generalisable than the somewhat more particular LAION case. It will be interesting to see how the Munich Regional Court positions itself on the key legal issues – especially in comparison to the Hamburg courts. What appears certain, however, is that the legal issues will ultimately be clarified by the Federal Court of Justice and the European Court of Justice.

2. Outlook

The free use of copyrighted works for the purpose of AI training has long been a source of displeasure for rights holders. After all, the quality of the training data, often created by human creativity, is the basis for the performance of an AI model, which then competes with the creators.

“AI lawsuits”, such as the LAION case before the Hamburg courts and the GEMA lawsuit against OpenAI, are therefore likely to become more frequent. From a legal perspective, they are as exciting as they are challenging, as a number of unresolved legal issues need to be resolved, and the individual proceedings often have their own peculiarities. In any case, it is very interesting to see that, in addition to individuals and media companies, collecting societies are now also acting as plaintiffs. According to GEMA, it is already considering taking action against other providers of AI systems.