A Munich court has ruled against AI music generation platform Suno in a copyright lawsuit filed by German collecting society GEMA, finding that the company infringed protected works while training and operating its artificial intelligence (AI) models. The decision orders Suno to disclose its revenue and pay damages, marking a significant legal test for generative AI in music.
Suno stated it disagrees with the ruling and is evaluating an appeal.
Allegations and the fair use question
GEMA brought the case alleging that Suno used copyrighted compositions from its members to train its models without authorization. The collecting society also argued that Suno’s technology could generate audio outputs substantially similar to existing musical works.
The dispute raised the issue of whether training AI systems on copyrighted material in the United States, where Suno’s models were allegedly trained, could be shielded by the US fair use doctrine.
Court’s reasoning on US and German law
In a summary of its decision, the Munich Regional Court determined that copies of musical works were made during the training process in the US, meaning US copyright law applied to those acts. The court distinguished this case from two US proceedings involving AI companies Anthropic and Meta, where certain uses of copyrighted material for AI training were found to qualify as fair use. The key difference, according to the German court, was that Suno’s systems could produce outputs containing elements substantially similar to the original works after simple, open-ended prompts.
The court also held Suno liable under German copyright law for storing its AI models on servers in Germany and for the reproduction and public performance of allegedly infringing soundalike recordings.
Suno’s response
“We built Suno around a core belief: to empower everyone to experience the joy of making music,” a Suno spokesperson said. “Our tools give people the ability to create new songs, whether they are top artists, product developers, songwriters using our tools in their workflows or everyday music fans. From the beginning, we trained our models to create new songs, not reproduce existing ones, and built protections into our platform. We disagree with today’s ruling – which rests on a fundamental mischaracterisation of how Suno’s technology works, how it is used and how US law applies – and are evaluating all available options, including an appeal.”
GEMA’s reaction
Kai Welp, GEMA’s general counsel, said the case underscored the need for further regulation of AI systems in Europe. “It is astonishing that the AI systems apparently store almost complete works to a considerable extent,” Welp said. “We were able to prove this with our methods and the recent scientific literature also indicates that this is only the tip of the iceberg. It is also very pleasing how the legal situation in the USA is assessed, and that we can sue in Munich regardless of where the training takes place. This is a huge step for the creatives. However, in order to effectively enforce our rights and prevent providers from moving to copyright-hostile countries, we will also need the support of the European legislator.”
Wider context and next steps
The ruling arrives as Suno faces broader examination over its use of copyrighted music in AI development and pursues licensing agreements with music industry stakeholders. The case is expected to remain subject to further legal proceedings if Suno proceeds with an appeal.
