In this case, GEMA, the German collecting society representing composers, lyricists and music publishers, is suing OpenAI LLC and its European subsidiary OpenAI Ireland Ltd. GEMA claims that OpenAI trained its language models, including ChatGPT, using copyrighted song lyrics from GEMA’s repertoire without authorization or remuneration.
According to GEMA, ChatGPT was able to reproduce substantial parts of protected lyrics when prompted, indicating that the training involved use of copyrighted material. The claims include unauthorized reproduction of works, as well as and making those works publicly available, as well as unauthorised adaptations (“hallucinations”) and infringements of moral rights. GEMA also relies on its formal opt-out from the text and data mining (TDM) exception, asserting that this opt-out excludes its repertoire from AI training without a licence.
After Gema was largely successful in its lawsuit before the Munich Regional Court and was able to prevail, particularly on the key copyright claims, the case is now pending before the appellate court, the Munich Higher Regional Court.