OpenAI Loses Landmark Song Lyrics Copyright Case In German Court

OpenAI has lost a major copyright case in Germany after a Munich court ruled that the company’s ChatGPT language model used protected song lyrics without paying royalties, violating German authors' rights laws.

The lawsuit was brought by GEMA, Germany’s leading authors’ rights organization, which argued that large language model (LLM) developers like OpenAI must pay licensing fees when their systems use copyrighted works, just as streaming platforms or online publishers do.

Judge Elke Schwager of the Munich District Court I found that OpenAI had unlawfully reproduced song lyrics and ordered the company to pay damages, though the exact amount was not disclosed. The ruling can still be appealed.

“We do not agree with the verdict and are examining further steps,” OpenAI said in a statement, adding that it respects intellectual property and is in talks with rights organizations worldwide.

The Case at a Glance

GEMA cited nine well-known German songs as evidence, including “Männer” by Herbert Grönemeyer, “In der Weihnachtsbäckerei” by Rolf Zuckowski, and “Atemlos” by Kristina Bach and Helene Fischer.

The court rejected OpenAI’s argument that ChatGPT doesn’t store or reproduce specific copyrighted material, but only “learns” patterns during training. The judge concluded that the reproduction of lyrics could not be coincidental, given their complexity and structure.

“Anyone who creates something using others’ content must pay or obtain permission,” Judge Schwager said, calling it surprising that a company capable of building such advanced technology had ignored a “clear legal situation.”

Broader Implications for AI and Copyright

The verdict could have sweeping implications for how AI models are trained and licensed in Europe, where copyright laws are closely aligned across member states.

GEMA welcomed the decision, saying it hopes the ruling will lead to constructive negotiations with AI companies over fair compensation. “The goal is not to remove technology from the market, but to ensure authors are properly paid,” said GEMA lawyer Kai Welp.

Media groups and journalists’ unions also hailed the ruling as a victory for creators. Mika Beuster, head of the German Journalists’ Association (DJV), called it “a partial victory for authors’ rights,” adding, “The training of AI models is intellectual property theft.”

OpenAI’s case marks one of the first major European rulings to directly address how copyright applies to AI training data — and could set a precedent for similar lawsuits worldwide.