A German Court Just Ruled That Training AI on Music Without a License Is Copyright Infringement

On July 31, 2026, the Munich Regional Court ruled against Suno, one of the biggest AI music generators in the world. Germany’s music rights organization GEMA brought the case. It won on almost every point. The ruling is the first in Europe to say clearly that AI companies need a license before using copyrighted music to train their models.

This is not a settlement. It is a binding court judgment.

What the Court Actually Decided

The case was heard by the 42nd Civil Chamber of the Munich Regional Court, case number 42 O 763/25, under presiding judge Elke Schwager. GEMA filed the lawsuit in January 2025. The original judgment date was June 12, then postponed to July 31.

The court found two separate violations.

First, storing protected songs inside Suno’s model infringes the reproduction right. Second, serving outputs built on those songs to users infringes the right of making works available to the public. Both violations were confirmed. Both carry consequences.

Suno must now stop the infringing uses, disclose its revenues to GEMA so damages can be calculated, and pay damages in an amount still to be determined. The court also granted GEMA an injunction covering Suno’s v3.5 and v4 models specifically.

The Six Songs That Sank Suno’s Defense

GEMA did not pick obscure tracks. The six songs used as evidence were chosen deliberately because they are immediately recognizable to any listener.

The test catalog: “Atemlos durch die Nacht,” “Forever Young” by Alphaville, “Big in Japan,” “Mambo No. 5” by Lou Bega, “Daddy Cool” by Boney M, and “Rasputin” by Boney M.

In the March hearing, GEMA’s legal team played the originals and Suno’s outputs back to back in the courtroom. The melodies matched. That demonstration was central to the ruling.

The court found that all six works are reproducibly contained in Suno’s models and recognizable in what those models produced. That finding means the model did not just learn musical patterns from these songs. It memorized them.

How Suno Got the Songs

The stream-ripping finding is the detail that carries the most weight going forward.

Judge Schwager stated after reading the verdict that it was undisputed that Suno pulled songs from the internet using stream-ripping techniques, including circumventing YouTube’s rolling cipher, a technical protection designed to prevent downloading. Suno did not dispute that the six works had been included in its training data.

The model was trained on more than two million songs scraped from the web. The court heard that number and was not sympathetic to arguments that the scale of the training data made individual song reproduction unlikely.

Three Defenses Suno Tried. All Three Failed.

Suno contested the ruling on multiple legal grounds. The court rejected all of them.

It rejected a US fair use defense. German copyright law does not recognize fair use in the same way US law does, and the court was not persuaded that US legal principles applied to conduct on German servers.

It rejected Germany’s text-and-data-mining exception. That exception exists in German law for non-commercial research purposes. The court ruled Suno’s commercial use did not qualify.

It put liability on Suno, not on users. That third point matters most for the industry. The court reasoned that Suno operated the models, selected the works as training data, and is responsible for the architecture and the memorization. Users typing prompts are not liable. Suno is.

What GEMA Said After the Ruling

GEMA CEO Tobias Holzmüller called it “a verdict of global significance.” He told Reuters that the ruling sends a clear message that AI companies cannot commercially exploit copyrighted music without compensating creators.

German Culture Minister Wolfram Weimer welcomed the decision, describing it as an important signal for protecting creators’ rights while establishing clear rules for AI innovation.

Holzmüller was also specific about GEMA’s goal. The organization does not want to eliminate AI music tools. It wants licensing negotiations. The ruling gives GEMA the leverage to force those negotiations.

Suno’s Response

Suno said it disagrees with the ruling and is evaluating all available options, including an appeal. The company’s full statement: “We built Suno around a core belief: to empower everyone to experience the joy of making music. Our tools give people the ability to create new songs, whether they are top artists, product developers, songwriters using our tools in their workflows or everyday music fans. From the beginning, we trained our models to create new songs, not reproduce existing ones, and built protections into our platform. We disagree with today’s ruling, which rests on a fundamental mischaracterization of how Suno’s technology works, how it is used and how U.S. law applies, and are evaluating all available options, including an appeal.”

The appeal is possible. German first-instance judgments can be appealed to a higher regional court. But the ruling is provisionally enforceable, meaning GEMA can seek to enforce the injunction while an appeal proceeds.

This Is Not the First Time GEMA Has Won

Nine months before the Suno ruling, GEMA won a separate copyright case against OpenAI in November 2025, which was found to have trained its models on unlicensed lyrics. GEMA was awarded unspecified damages in that case too.

Two major AI copyright victories in under a year from the same German rights organization is a pattern, not a coincidence. GEMA has a strategy, and it is working.

The Broader Legal Pressure on Suno

The German ruling is not the only legal problem Suno AI is managing.

In the United States, the Recording Industry Association of America filed copyright lawsuits on behalf of Sony Music Entertainment, Universal Music Group, and Warner Records, all making similar allegations about unauthorized use of copyrighted recordings for AI training. More than 1,800 artists are supporting class-action lawsuits against Suno and rival Udio.

Sony has also filed a separate suit against Udio covering more than 30,000 songs. And Warner Music reached a licensing settlement with Suno, under which Suno agreed to launch licensed AI models in 2026 to replace its current platform.

The direction of the legal landscape is now clear. AI music tools that trained on scraped music without licenses are facing binding rulings, growing damages exposure, and forced licensing negotiations on multiple continents.

What This Ruling Changes

Before July 31, 2026, no European court had issued a binding judgment on whether AI training on copyrighted music requires a license. That question is now answered in Germany, and the answer is yes.

The ruling covers both sides of the process. Training the model on protected music is infringement. Generating outputs that reproduce recognizable elements of those songs is also infringement. Both require licenses. Both carry damages.

For every AI music company operating in Europe, this ruling changes the risk calculation immediately. The “train on everything and deal with lawsuits later” approach now has a specific dollar figure attached, even if that figure has not been quantified yet in Suno’s case.

The era of training AI on music without asking permission is closing. Licensing is becoming the price of doing business in this space.