German court rules Suno AI violated copyrights, orders damages in landmark case

A Munich court has ruled that Suno, a US-based artificial intelligence company that generates songs from text prompts, violated copyright law by training its models on protected music without permission or payment. The verdict, handed down Friday, orders the company to disclose revenues gained through the infringement and to pay damages—the amount still to be determined. Suno has said it disagrees with the decision and is considering an appeal.

The lawsuit was filed in January 2025 by GEMA, Germany's state-mandated music licensing agency, which represents more than 95,000 composers and songwriters domestically and over 2 million rights holders worldwide. GEMA's central argument was straightforward: Suno trained its AI system on copyrighted songs without obtaining licenses or compensating the artists, songwriters, and publishers whose work made that training possible. The company then built a commercial platform around that unlicensed foundation, generating revenue while the original creators saw nothing.

What makes this case significant is how GEMA proved its point. During the proceedings, the agency prompted Suno's system to create new songs using the original lyrics of well-known tracks—"Forever Young" by Alphaville, "Mambo No. 5" by Lou Bega, "Daddy Cool" by Boney M., and others from GEMA's catalog. The plaintiffs provided only the lyrics, the desired musical style, and the song title, without specifying melody, rhythm, harmony, or arrangement. Yet the AI produced songs strikingly similar to the originals. Suno acknowledged that these songs were indeed part of its training data. The company argues, as does much of the AI industry, that ingesting data for training purposes falls under "fair use" in the United States and the "text and data mining exception" in European law. But whether such unlicensed training is actually permitted under copyright law remains fiercely contested globally, and courts are now examining whether it causes unfair economic harm to original creators.

Tobias Holzmüller, GEMA's CEO, called the verdict "of global significance." He emphasized that the goal is not to eliminate Suno but to force it to the negotiating table as an equal partner—something he says has been impossible until now. He describes Suno as "by far the largest and commercially most successful of all AI music models" and notes that it pursues a far more aggressive business strategy than competitors. The scale of the operation underscores why this matters: across Suno's global platform, users generate roughly 7 million songs per day, with about 75,000 of them reaching streaming services daily.

The case is not the first of its kind. Last year, GEMA won a similar lawsuit against OpenAI in the same court, proving that the company had unlawfully trained its models on copyrighted song lyrics and reproduced them through ChatGPT without authorization. OpenAI was ordered to stop the practice, disclose information, and pay damages—though it is appealing. The current Suno case differs in that it focuses on entire musical compositions rather than lyrics alone.

Martin Senftleben, a professor of intellectual property law at Amsterdam Law School, notes that the lawsuit's significance lies partly in who brought it. Because GEMA is a collecting society rather than a private company or record label, it can represent a far broader swath of creators—not just famous artists but ordinary songwriters and composers who are GEMA members. This means potential compensation could reach a much wider group of rights holders than a single-label lawsuit might.

Holzmüller rejected the common argument that AI learns from music the way humans do. "These are concepts designed for the human brain, not for a machine that can digest millions of sound files in seconds," he said. The sheer volume of data consumed and processed by these systems bears no comparison to human learning. To frame it otherwise, he argued, is neither logical nor moral. His concern is existential for the music industry: if AI-generated music remains free or cheap while human-created music carries licensing costs, the market will inevitably favor the machines. The only sustainable path forward, in his view, is to attach a price to AI output—licensing fees that would make artificial and human-created music economically equivalent. Then, he believes, people will choose human music for its irreplaceable qualities. Without that correction, he warns, AI will push human creators out of markets where they have long made their living—film scores, social media soundtracks, and countless other spaces. The court's decision is a first step toward that reckoning, but the real test will be whether it holds on appeal and whether other courts around the world follow suit.