A noteworthy European legal case concerning AI music copyright is approaching a significant conclusion on July 31, with the Munich Regional Court poised to issue a ruling in the lawsuit initiated by GEMA, Germany's music collecting agency, against Suno AI, a U.S. firm known for generating songs from textual prompts.
GEMA filed the lawsuit in January 2025, asserting that Suno trained its AI systems using copyrighted music without securing the necessary licenses or compensating the rightful owners of those works. This case is attracting considerable attention as it marks one of the earliest instances globally where traditional copyright laws are being evaluated in the context of AI-generated music and its training process.
The proceedings will potentially clarify whether companies that produce AI music must obtain licenses prior to training their models on copyrighted tracks and whether AI compositions that bear similarities to existing songs infringe copyright laws.
"Our intention isn’t to obliterate Suno from existence but to initiate licensing dialogues on equal footing, which has not been feasible with Suno up until now," expressed GEMA CEO Tobias Holzmüller. He labeled Suno as "arguably the largest and most commercially successful AI music model," contrasting its aggressive business tactics with those of other companies.
The resolution of this lawsuit is tied to a fundamental issue: How should artists receive compensation when their music is utilized to train generative AI models?
Last year, GEMA achieved a significant victory against OpenAI in a similar court case, where the court determined that OpenAI unlawfully used copyrighted song lyrics to train its AI models and reproduce them through ChatGPT without authorization. OpenAI was instructed to stop reproducing the copyrighted works, share relevant information, and pay damages. However, the case remains unsettled as the company has appealed the ruling.
In this current lawsuit, GEMA's focus has shifted from song lyrics to entire music compositions. According to Martin Senftleben, a professor of intellectual property law at Amsterdam Law School, this case is significant not only because it’s initiated by a music collecting society rather than a private entity, but also due to its inclusivity. GEMA represents over 95,000 creators, including composers and songwriters, providing a stronger platform for rights holders to seek proper remuneration against AI-generated outputs.
GEMA contends that Suno has used copyrighted materials without permission, profiting from the work of songwriters, composers, and music publishers it represents. Suno's platform allows users to create songs by simply inputting text-based prompts about desired moods, styles, or genres, yielding complete tracks in under a minute, often reminiscent of existing copyrighted works.
During court proceedings, GEMA illustrated this by prompting Suno AI to create songs based on original lyrics from popular tracks like "Forever Young" by Alphaville and "Mambo No. 5" by Lou Bega, demonstrating striking similarities between the AI-generated music and the originals.
Suno has acknowledged that several of the songs mentioned in GEMA's lawsuit were employed in training its AI model. The company argues that the data acquisition for AI training falls under the "fair use" doctrine in the U.S. and the "text and data mining exception" in the EU, although the legality of such practices remains highly contested across various jurisdictions.
Senftleben noted that this lawsuit is distinctive in that GEMA’s focus is on the end product of the AI systems rather than solely on the training process. By proving that these AI applications generate music closely aligned with copyrighted compositions, GEMA aims to highlight potential market damages at the output stage.
While AI developers assert that their models learn from existing music just as humans do, rights holders believe these systems merely replicate and exploit their works commercially. Holzmüller critiques the analogy, stating that the cognitive processes of a human musician cannot be equated with the capabilities of AI, which rapidly processes vast amounts of data.
Suno's users generate around 7 million tracks daily, with approximately 75,000 of those being published on streaming platforms. Holzmüller warns that this volume is unsustainable and harmful to the music ecosystem. He advocates for establishing a pricing mechanism for AI-generated music to ensure it does not undermine human-created works.
He believes that if the licensing costs for AI-generated music rise to match those of traditional songs, people will likely continue to choose human-made music for its unique value.
