A Munich regional court ruled that the U.S.-based AI music company Suno violated German copyright law by using protected works [1].
The decision marks a significant legal setback for generative AI firms that rely on vast datasets of existing music to train their models. By ruling that the use of copyrighted music without a license constitutes infringement, the court establishes a strict precedent for how intellectual property is handled within the European Union's legal framework [1], [2].
The legal challenge was brought by GEMA, Germany’s music collection society, which represents the interests of composers and songwriters [3]. GEMA said that Suno integrated copyrighted music into its AI-generated songs without obtaining the necessary permissions or paying royalties to the original creators [3].
The court found that Suno's practice of utilizing these works to power its music service infringed upon the rights of the copyright holders [1], [2]. This ruling focuses on the unauthorized use of music during the training and generation process, a central point of contention in global debates over AI and art [2].
Suno is currently considering an appeal of the decision [1]. The company has previously defended its technology as transformative, though the Munich court did not find that defense sufficient to override German copyright statutes [1], [2].
The ruling comes as regulators and artists worldwide seek to define the boundaries of "fair use" in the age of artificial intelligence. While some jurisdictions are still debating whether AI training constitutes a new form of creativity or a digital theft, the Munich court has sided with the rights holders [2].
“The court ruled that Suno violated German copyright law by using copyrighted works.”
This ruling reinforces the legal strength of music collection societies like GEMA and suggests that AI companies cannot rely on broad interpretations of innovation to bypass licensing fees in Germany. It creates a potential roadmap for other European nations to pursue similar copyright infringement claims, potentially forcing AI firms to negotiate expensive licensing deals or limit their training data to public-domain works.
