DRULES AI
Suno has struck a licensing deal with BMG, paying for the song catalog it scraped to train its models. The move comes days after a German court ruled against the company, declaring unauthorized use of protected melodies illegal and ordering disclosure of financials to calculate damages.
⚖️ The German Court Hammer
The ruling marks the first major legal strike against an AI music generator's training practices in Europe. Judges found Suno reproduced protected elements without permission, opening the door for substantial penalties. Suno is expected to appeal, but the order to open books could expose exactly how much copyrighted material was ingested during development.
Industry watchers note this isn't isolated. With multiple lawsuits pending from labels and publishers, the BMG agreement signals a pivot from pure legal defense to selective settlements that keep the lights on.
💼 BMG Gets Paid, Creators Get a Cut?
Under the deal, BMG songwriters will receive royalties when Suno-generated tracks mirror their work. It's a pragmatic workaround that gives Suno legal cover while the majors continue fencing off AI content on streaming platforms like Spotify.
Critics argue this rewards the very scraping behavior the court condemned. For professional creators using Suno, it raises questions about output ownership and potential future claims if generated tracks start charting.
📈 What It Means for the Ecosystem
Google's Lyria API suddenly looks cleaner from a licensing perspective, while Udio and smaller tools face similar scrutiny. The deal doesn't resolve the broader copyright war but sets a template: pay-to-train partnerships may become table stakes for platforms that want to scale.
Community sentiment on X is split between relief that Suno survives and frustration that user trust took another hit after recent ToS changes and watermarking mandates.
Bottom line: Suno is buying legitimacy one label at a time while appeals buy time, but the German precedent just made training data the most expensive ingredient in AI music.