Overview
- Landgericht München I largely sided with GEMA, finding that ChatGPT memorized and reproduced nine specified song texts in violation of German copyright law.
- The court ordered OpenAI to stop storing and outputting those lyrics, to pay damages, and to disclose information on use and related revenues.
- Judges deemed near-verbatim outputs evidence of memorization and rejected defenses based on probabilistic synthesis or user responsibility for prompts.
- OpenAI said it disagrees with the decision and is considering further steps, with appeals and a potential referral to the European Court of Justice anticipated.
- Observers say the case could reshape licensing expectations for AI training across music, journalism, literature, photography and other creative sectors.