German Court: AI Can't Treat Internet as 'Self-Service' for Lyrics

A German court has ruled that OpenAI's ChatGPT chatbot violated copyright laws by using lyrics from German songs without authorization, marking a landmark decision in Europe's evolving legal landscape around artificial intelligence. The Munich Regional Court sided with GEMA, Germany's music rights society, which argued that OpenAI's AI models were trained on protected content from its repertoire of approximately 100,000 members, including prominent artists like Herbert Groenemeyer, as Reuters reported. The ruling requires OpenAI to pay undisclosed damages and obtain licenses for the use of copyrighted material, as the Guardian reported.

The court found that ChatGPT's language models "memorize" and reproduce lyrics in response to user prompts, infringing on reproduction rights, according to a Music Business Worldwide report. This applies to nine German songs, including Groenemeyer's 1984 hit Männer and Helene Fischer's Atemlos Durch die Nacht, which were central to GEMA's case, as the Guardian reported. GEMA's CEO, Dr. Tobias Holzmüller, emphasized that the verdict establishes a precedent: "The internet is not a self-service store, and human creative achievements are not free templates," as Music Week reported. The court rejected OpenAI's argument that its AI's training on public data qualified as fair use under text and data mining exceptions, as WebPronews reported.

OpenAI, which maintains that its models learn patterns rather than store specific content, has announced it will consider an appeal. The company stated the ruling applies to a "limited set of lyrics" and does not affect its broader operations in Germany, as the Guardian reported. However, legal experts note the decision could influence ongoing lawsuits against OpenAI in the U.S. and other European cases, such as GEMA's pending action against AI music generator Suno, as Music Business Worldwide reported.

The ruling has broader implications for the AI industry. It underscores the need for licensing agreements between tech firms and creators, a model GEMA has proposed for AI tool providers, as Music Business Worldwide reported. Judge Elke Schwager, presiding over the case, stressed that "both the memorization in the language models and the reproduction of the song lyrics constitute an infringement of copyright exploitation rights," as Juve Patent reported. This aligns with Germany's strict copyright framework, which prioritizes creators' rights over technological innovation in commercial contexts, as WebPronews reported.

For the music industry, the verdict signals a potential shift toward fair compensation for artists in the AI era. GEMA's legal counsel, Kai Welp, called it a "milestone" for European copyright law, asserting that even "tech giants must obtain licenses for the use of intellectual property," as Music Business Worldwide reported. Meanwhile, OpenAI faces growing scrutiny as it navigates similar litigation in the U.S., where authors and media groups allege its models were trained on their works without permission, as Mezha reported.

The case also highlights tensions between AI development and intellectual property rights globally. As AI systems become more sophisticated, courts are increasingly tasked with defining boundaries for data usage. This ruling, the first of its kind in Europe, could set a template for future disputes, particularly in jurisdictions with robust copyright protections, as WebPronews reported.