OpenAI infringes song copyright, according to a German court

This is a watershed decision for a music and literary creation sector that needs stricter regulation in Europe, in the face of tech giants’ control over its production.

France Télévisions – Cultural Editorial

Published


Updated


Reading time: 4 minutes

A German court ruled that the creators of ChatGPT exploited musical works without compensating their authors. (WLADIMIR BULGAR / SCIENCE PHOTO LIBRARY / WBU / AFP)

A German court ruled that the creators of ChatGPT exploited musical works without compensating their authors. (WLADIMIR BULGAR / SCIENCE PHOTO LIBRARY / WBU / AFP)

Players in the German music industry scored their first legal victory against OpenAI on Tuesday, November 11, with a court finding that the American artificial intelligence giant had violated copyrights related to song texts.

THAT “linguistic model” used by OpenAI as well as “reproduction of song texts in chatbot results” is “violation of copyright protected exploitation rights”, a court in Munich, Germany ruled, summarizing its decision in a press release.

OpenAI reacted by revealing it “contradictory”a company spokesperson told AFP. “We are studying possible next steps”he added, further stressing that the decision only concerns the texts that appear in the complaint, in particular the hits of stars in Germany, Helene Fischer and Herbert Grönemeyer. “This decision is just words and has no impact on the millions of people, developers in Germany who use our technology every day”he added.

Although the court granted most of the requests from the German Society for the Collective Management of Music Copyright (Echo), which sought damages, the court did not say how this could be calculated. Echo, which represents around 100,000 players in the music industry in Germany, filed a complaint in November 2024, accusing OpenAI of using song lyrics to train its AI models and then returning them to users, without a license or remuneration from the authors.

The trial is the first in Europe, according to Echo, which refuses to allow human work to serve as free raw material for the AI ​​giant, and for anyone else. “creator’s livelihood” at stake. Organizational demands “found, either due to the reproduction of the text in the linguistic model or its reproduction in the results”estimate the court.

On the first point, “indirect perception” of this work is “sufficient to constitute a reproduction”court estimates based on the case law of the Court of the European Union (CJEU). And in response, the chat robot ChatGPT has it “Making the lyrics of the disputed songs accessible to the public in an unauthorized manner”.

The decision concerns song texts by nine famous German authors. OpenAI has rejected the accusations, arguing that its models do not store individual data but rather “reflects what they learned based on the training data set.”

OpenAI, whose ChatGPT claims about 700 million weekly users, is considered one of the world leaders in artificial intelligence. These giants’ control over musical and literary works is often criticized by industry players, who demand stronger regulation, in particular through European legislation (the AI ​​Act), to gain transparency over the data used and guarantee their revenues. For the law firm Raue, which represents Gema, the Munich decision “provides legal certainty for creators, music publishers and platforms across Europe and could have an impact far beyond Germany.”

According to the international music publishing organization, the Brussels-based International Confederation of Music Publishers (ICMP), tech giants have sucked up “all world music” without respecting copyright, according to a study published Sept. 9 in music publication Billboard (link in English).