The judgement is not final and OpenAI has said that it is considering next steps.

OpenAI has violated German copyright laws by reproducing song lyrics by popular artists, the Munich Regional Court has ruled today (11 November).

GEMA, a German music rights organisation representing more than 100,000 composers, songwriters and publishers, filed a lawsuit against OpenAI in 2024. The case references nine songs by popular German artists, and GEMA claims that its lyrics were memorised by the large language models (LLM) and reproduced when simple prompts were put into the chatbot.

“Both the memorisation in the language models and the reproduction of the lyrics in the outputs of the chatbot would interfere with the copyright exploitation rights,” a translated press release put out by the Regional Court read.

OpenAI claimed that its LLMs did not store or copy specific training data. It said that since the chatbot’s outputs are dependent on prompts put in by users, it is not their responsibility, but rather the users’, who generate reproduced outputs.

The court, however, found that the disputed lyrics are “reproducibly” contained in OpenAI’s GPT 4 and 4o models, after being ‘memorised’ for training. It added that the defendants also unjustifiably duplicated the lyrics and made them publicly available by allowing them to be outputted by the chatbot.

“The defendants, not the users, are responsible for this. The outputs were generated by simple prompts,” the court said. Although, the judgement is not final and can be appealed.

“The internet is not a self-service store, and human creative achievements are not free templates,” said GEMA CEO Tobias Holzmueller, in a statement to the press.

An OpenAI spokesperson, however, told SiliconRepublic.com that the company is considering next steps. “The decision is for a limited set of lyrics and does not impact the millions of people, businesses and developers in Germany that use our technology every day.

“We respect the rights of creators and content owners and are having productive conversations with many organisations around the world, so that they can also benefit from the opportunities of this technology.”

GEMA has also filed a similar lawsuit this year against Suno AI, an AI music generator.

The ruling comes as copyright laws worldwide have come to the fore in the generative AI (genAI) age where LLMs commonly consume copyrighted material to train and hone their results.

Just last week, Getty lost a major lawsuit in the UK against Stability AI after the country’s High Court ruled that the AI company’s stable diffusion model weights did not contain copies of Getty’s copyrighted works.

And earlier this year, a US court sided with AI giant Anthropic, denying a motion filed by mega music publishers Universal Music Group, Concord and Capitol CMG to stop the start-up from using their song lyrics to train its AI models.

While the complaint referenced 500 songs, the plaintiffs said that the proposed injunction would extend to all of their works. Anthropic pushed back, claiming that excluding an undefined number of material would make training its models “virtually impossible”.

Also earlier this year, Anthropic lost a major copyright battle against book publishers over illegally pirating their work to train its models. Although the company wound up having to pay $1.5bn to settle the case, the court ultimately said that the AI giant is protected under fair use when it came to using books to train Claude.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Updated, 1.40pm, 11 November 2025: This article was updated to reflect that SiliconRepublic.com received a statement from OpenAI.