Suno Loses Landmark Copyright Case in Germany

The Munich Regional Court delivered a landmark defeat to Suno Friday, ruling the AI music platform violated copyright law by training its models on GEMA's repertoire without licenses or compensation to rights holders.

⚖️ What the Court Decided

Judges found Suno infringed both in the training phase and through storage and reproduction of protected works. The case centered on six specific songs including "Daddy Cool," "Rasputin," "Forever Young," and "Mambo No. 5." Evidence showed the model had reproducibly memorized elements of these tracks. GEMA, Germany's collecting society, argued successfully that commercial AI systems cannot scrape and train on copyrighted catalogs for free.

Leaked data referenced in coverage revealed Suno's training pulled over 113,000 hours from YouTube Music, plus tens of thousands from Pond5 and Deezer. The ruling mandates licenses for both training data and generated outputs, extending beyond simple output similarity to internal model states.

💰 Orders and Suno's Stance

Suno must disclose revenue tied to the infringement and pay yet-to-be-determined damages. While not yet legally binding pending appeal, the decision carries global weight. Suno responded sharply: “Our tools give people the ability to create new songs... We disagree with today’s ruling—which rests on a fundamental mischaracterization of how Suno’s technology works—and are evaluating all available options, including an appeal.”

This follows GEMA's prior win against OpenAI for unauthorized lyric reproduction, underscoring Europe's aggressive stance on AI training data. Major labels have also sued Suno and Udio in the US, making licensing the central battleground.

🌐 Impact on AI Music Platforms

For creators using Suno professionally, the verdict introduces immediate uncertainty around commercial releases. Platforms may face higher costs as they negotiate blanket licenses with collecting societies, potentially raising subscription prices or limiting catalog access. Udio, Google Lyria, and others now operate under heightened legal risk, accelerating talks for provenance-tracking tools and licensed datasets.

Industry watchers note the ruling links training data directly to infringement, rejecting fair use arguments common in the US. This could chill rapid iteration in generative audio and push developers toward ethically sourced or synthetic training data. Early reactions on X highlight fears of slowed innovation but praise clearer rules for rights holders.