Skip to main content

German Court Rules AI Music Generator Suno Violates Copyright

In a landmark decision that could send ripples through the AI industry, a Munich court has ruled against the popular AI music generator Suno, finding it guilty of copyright infringement. The court rejected Suno's fair use defense, marking a significant victory for music rights holders and setting a precedent that may reshape how AI companies approach training data.

The lawsuit, initiated by the German music copyright organization GEMA, centered on six well-known songs, including Kristina Bach's "Atemlos durch die Nacht" and the classic "Rasputin." After a thorough review, the court determined that versions 3.5 and 4 of Suno's model could accurately reproduce original elements from these tracks. This phenomenon, known as "memorization," occurs when an AI model doesn't just learn general patterns but stores specific original content that can be retrieved later as generated output.

Suno had argued that it only stored mathematical patterns and that the outputs were the result of user prompts. But the court wasn't buying it. The judges pointed out that Suno selected its training data and designed its model architecture, so it bears direct responsibility for the infringing outputs—not the users who provided open-ended prompts. The court also noted that Germany's text and data mining exception doesn't apply to this kind of "memorized copying."

What's more, the court even exercised jurisdiction over training activities that occurred in the United States, applying U.S. law to determine that the fair use doctrine doesn't protect Suno. Unlike some U.S. cases where AI outputs were deemed transformative, Suno's model can generate results highly similar to original works with just a simple input.

During the trial, another key detail emerged: Suno used streaming scraping technology to extract music from YouTube, bypassing technical protection measures designed to prevent downloads. This revelation adds another layer of concern about the company's data sourcing practices.

This ruling is still under review and not yet final, but its implications are profound. It sends a clear message to AI companies: you can't hide behind user prompts or fair use when your model reproduces copyrighted material. The decision could force the entire industry to rethink how it collects training data and ensure compliance with copyright laws.

For musicians and rights holders, this is a hopeful sign that their work is protected in the age of AI. For AI developers, it's a wake-up call to be more transparent and responsible about their training methods. As the case continues to unfold, all eyes will be on how this ruling influences future litigation and legislation around AI and copyright.

Key Points

  • Court Ruling: Munich court finds Suno guilty of copyright infringement in both training and output.
  • Fair Use Rejected: Suno's defense of fair use was dismissed, both under German and U.S. law.
  • Memorization Issue: The court highlighted that Suno's models could reproduce original song elements, not just learn patterns.
  • Jurisdiction: The court asserted jurisdiction over U.S. training activities, applying U.S. fair use standards.
  • Data Sourcing: Suno was found to have used streaming scraping to bypass YouTube's download protections.
  • Industry Impact: The decision could reshape AI training data practices and copyright compliance across the industry.