Source: Mondaq

Mondaq details a significant judgment delivered by the District Court of Munich I on November 11, 2025, in the case of GEMA (the German musical collecting society) versus OpenAI. The court ruled that OpenAI infringed GEMA's copyrights by allowing its ChatGPT model to reproduce the lyrics of protected German songs. The court found that the "memorization" of these lyrics within the model's parameters constituted an unauthorized reproduction under copyright law, and the subsequent outputting of these lyrics to users amounted to an unauthorized "making available to the public."

Crucially, the court rejected OpenAI's defense relying on the Text and Data Mining (TDM) exception, stating that the exception did not apply to the generation of competing creative content. The ruling imposes liability on AI providers for the content their models generate and store, suggesting that companies must either license the training data or implement strict filters to prevent the regurgitation of protected works. This decision has profound implications for the AI industry in Europe, potentially mandating broad licensing agreements for copyrighted material used in LLM training.