🧔♂️ A friendly human may check it before it goes live. More news here
German court rules ChatGPT violated copyright by using song lyrics
A Munich court has ruled that OpenAI’s ChatGPT breached German copyright laws by generating lyrics from songs by musician Herbert Groenemeyer and others.
The case was brought by German music rights society GEMA, which said the AI was trained on protected content from nine German songs, including Groenemeyer’s “Maenner” and “Bochum.”
The court ordered OpenAI to pay damages for using copyrighted material, though the amount was not disclosed.
OpenAI argued that its models do not store or copy specific training data, but the court found that both the memorization and reproduction of lyrics by the chatbot violated copyright.
The ruling could influence how AI companies in Europe handle copyrighted materials.
OpenAI said it disagrees with the decision and is considering its next steps.
The case adds to global legal scrutiny of AI firms’ use of copyrighted works for training models.
🔗 Source: Reuters
🧠 Food for thought
Implications, context, and why it matters.
- The Munich Regional Court held OpenAI liable for damages and banned reproducing unlicensed German song lyrics or storing related data on infrastructure in Germany 1. It ordered publication of the judgment in a local newspaper at OpenAI’s expense, case 42 O 14139/24, with no appeal timetable in public materials 12.
- The order names GPT-4 and GPT-4o, a multimodal version; coverage of newer releases is disputed 2. OpenAI said the decision covers limited lyrics, does not affect the millions of people, businesses, and developers in Germany that use its technology, while weighing next steps 3. The court refused a referral on EU copyright law to the European Court of Justice, an EU process where courts ask its top court to interpret that law 1.
- The ruling says OpenAI needs a GEMA licence to train ChatGPT on German lyrics 2. It said model memorisation plus output reproduction infringe exploitation rights and fall outside the Text and Data Mining restriction under Section 44b of Germany’s copyright law 2.
- Vendors can build machine-readable opt-out tools 45. The Hamburg Regional Court suggested in obiter, a non-binding remark, that natural language terms and conditions could count if AI tools can interpret them 45. Publishers plus content platforms need tooling for metadata (structured tags about content), robots.txt files (a standard file that tells automated crawlers which pages they may access) or website terms, which creates a market for rights-management infrastructure that bridges human- and machine-readable formats that AI systems must honor before training 5.
Recent OpenAI developments
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.




