A Munich Regional Court ruled Friday that AI music generator Suno breached German and U.S. copyright laws by using protected works for training [1].
The decision marks a significant legal setback for generative AI companies that rely on massive datasets of existing creative works without obtaining licenses from rights holders. It establishes a judicial precedent in Europe regarding the legality of using copyrighted music to build commercial AI models.
The ruling followed a lawsuit filed by GEMA, the German music collecting society. The court said that Suno used, stored, and reproduced copyrighted musical works without permission [1], [2]. These actions violated both German and U.S. copyright statutes [1], [3].
As a result of the breach, the court ordered Suno to disclose any illicit revenue generated through the use of these protected works [1]. The court's determination focuses on the unauthorized nature of the data ingestion process used to train the AI's music generation capabilities [3].
Suno is a U.S.-based company that allows users to create songs via text prompts. GEMA said that the company's training methods bypassed the legal requirements for licensing music, effectively stealing the intellectual property of composers and songwriters [2], [4].
The ruling occurred on July 31, 2026 [1]. While the company had argued for the legality of its training processes, the Munich court rejected those claims, citing the clear infringement of existing copyright rules [1], [3].
“Suno breached German and U.S. copyright laws by using protected works for training”
This ruling reinforces the 'opt-in' or licensed model for AI training in Europe, signaling that the 'fair use' or 'text and data mining' defenses may not protect AI firms when they commercialize tools built on copyrighted art. By ordering the disclosure of illicit revenue, the court is creating a mechanism for calculating financial damages, which could lead to massive settlements or the forced retraining of AI models without infringing data.
