A Munich court has ruled that OpenAI violated German copyright law by training its AI models on lyrics from nine songs, including popular tracks by Groenemeyer, such as ‘Maenner’ and ‘Bochum.’ The presiding judge, Elke Schwager, ordered OpenAI to pay damages, although the exact amount was not disclosed. The case was brought by GEMA, a German music rights society representing composers, lyricists, and publishers. This decision marks another significant development in the global effort by artists to combat data scraping by AI systems.
GEMA’s legal advisor, Kai Welp, stated that the organization hopes to engage in discussions with OpenAI regarding the proper compensation of copyright holders. OpenAI had previously defended its position, asserting that its language models did not store or copy specific training data but instead learned from the entire dataset. The company argued that the output generated by the models was based on user inputs known as prompts, making the users rather than the company liable for any potential copyright violations.
However, the court found that both the memorization of language models and the reproduction of song lyrics in the chatbot’s outputs constituted copyright infringement. This ruling highlights the complexities of AI development and intellectual property law, as companies and content creators continue to navigate the legal landscape surrounding data usage and copyright enforcement.