German Court Rules Against OpenAI: ChatGPT Found Liable for Copyright Violations in Song Lyrics Case

In a landmark ruling with far-reaching implications for the global artificial intelligence industry, the Munich Regional Court has sided with GEMA, Germany’s leading copyright licensing organization, in its lawsuit against OpenAI, the developer of ChatGPT.

The court ruled that ChatGPT violated the copyrights of nine popular songs by using their lyrics without authorization, and that OpenAI may not use or reproduce any lyrics from artists registered with GEMA unless it first secures a proper licensing agreement.

The decision marks one of the first major legal precedents in Europe addressing the relationship between generative AI models and copyright law, and could influence how artificial intelligence systems are trained and deployed worldwide.

Court Sides with GEMA: ChatGPT Cannot Use Unlicensed Song Lyrics

The Munich court accepted GEMA’s core argument that ChatGPT’s automated reproduction of song lyrics constitutes an act of unauthorized duplication and distribution under German copyright law.

The ruling found that ChatGPT’s use of protected lyrics—both during model training and while generating responses for users—breaches intellectual property rights unless the company obtains an explicit license from the rights holders.

The court declared that:

“ChatGPT cannot use, store, or reproduce copyrighted song lyrics belonging to GEMA’s registered artists unless a valid licensing agreement is established.”

The decision also held OpenAI liable for any financial damages arising from these copyright infringements.

Furthermore, the court banned OpenAI from storing or reusing the disputed texts within its language models and dismissed both parties’ requests to escalate the case to the Court of Justice of the European Union (CJEU), keeping the verdict at the national level for now.

A Dispute Over Data and Creativity

GEMA, which represents more than 85,000 artists and composers, argued that ChatGPT’s training methods violated copyright law by copying and reproducing song lyrics without authorization.

According to GEMA’s legal team, the model’s use of these texts amounts to a form of large-scale, automated plagiarism, because each lyric reproduced—whether fully or partially—constitutes a direct copy of a creative work.

OpenAI, however, has denied all allegations, claiming that the AI model does not “store” copyrighted works in the traditional sense and that any generation of lyrics is based on statistical patterns, not memorized content.

GEMA countered that explanation, insisting that training on copyrighted material without a license still qualifies as reproduction, regardless of whether the output matches the original text verbatim.

Beyond Music: A Precedent for the AI Industry

Experts note that the decision could have implications far beyond song lyrics. Should higher courts uphold the ruling, the precedent may extend to other creative domains such as literature, journalism, photography, film scripts, and visual arts—essentially any medium used in AI training datasets.

Dr. Katrin Heine, a digital law specialist at Humboldt University, commented:

“If the Munich verdict stands, it may become the cornerstone of European AI copyright regulation. The outcome could affect every generative model that relies on human-created material.”

Analysts predict that the case will likely be appealed to a higher court, where the issue of what constitutes “use” or “reproduction” in AI training may be scrutinized more deeply.

The verdict could also force major AI developers—including OpenAI, Google, and Anthropic—to negotiate formal licensing agreements with collecting societies such as GEMA, PRS, and SACEM, to avoid further litigation.

A Turning Point for Generative AI Regulation

While GEMA has welcomed the decision as a victory for artists’ rights, the case raises complex questions about how copyright law should adapt to the age of machine learning.

If upheld, the Munich court’s reasoning could effectively make unlicensed data training illegal in Germany and across the EU, unless exceptions for research or “transformative use” are later defined by legislation or the European Court of Justice.

Legal scholars say that such rulings are likely to multiply across Europe as creators, publishers, and tech companies battle over how creative works are used in AI development.

As one copyright lawyer summarized:

“This decision goes far beyond song lyrics—it touches the very foundation of how AI learns. The line between innovation and infringement is being redrawn in real time.”