Today, the Munich Regional Court issued Europe’s first judgment finding that OpenAI infringed copyright by (i) memorising (song) texts and (ii) displaying parts thereof in the output. The judgment on behalf of music collecting society GEMA is groundbreaking on many topics:
“Both the memorisation in the language models and the reproduction of the song lyrics in the chatbot's outputs constitute infringements of copyright; not covered by any limitations, in particular [..] text and data mining.’
"Memorisation [...] occurs when the language models not only extract information from the training data set during training, but also completely adopt the training data in the parameters specified after training. [...] The song lyrics at issue are reproducibly defined in the models.”
"If not only information is extracted from training data during training, but works are also reproduced, this does not constitute text and data mining [..] In the case of reproductions in the model [..] the exploitation of the work is permanently impaired and the legitimate interests of the rights holders are thereby infringed.”
“[T]he defendants also unlawfully reproduced and made publicly available the song lyrics in question by reproducing the lyrics in the chatbot's outputs. The original elements of the song lyrics would always be recognisable in the outputs.”
“The defendants, and not the users, are responsible.”
The judgment, if upheld, will have far-reaching consequences – a very welcome clarification on copyright law in the era of GenAI.
Big congrats to GEMA, Robert Heine and his team for bringing this important precedent!
The unofficial courtesy translation of Court’s press release is attached.