AI Music Generators Can Infringe Copyright — Munich Court Rules Against Suno

On July 31, 2026, the Munich Regional Court handed down the first European ruling against an AI music generator. In GEMA v. Suno (Case No. 42 O 763/25), Judge Elke Schwager’s 42nd Civil Chamber found that Suno — a U.S.-based AI music platform — infringed copyright by training its models on six protected musical works without authorization, storing those works inside its model parameters, and generating outputs that reproduced them in recognizable form. The court rejected Suno’s fair use defense, ordered injunctive relief backed by fines of up to €250,000 per violation, and awarded damages and disclosure claims to GEMA, the German collecting society representing over 100,000 composers, lyricists, and music publishers.

For indie artists who use or distribute AI-generated music, this ruling carries immediate and practical consequences.

What the Court Actually Decided

The case centered on six musical works from GEMA’s repertoire, including Boney M’s “Rasputin,” Lou Bega’s “Mambo No. 5,” and Alphaville’s “Forever Young.” GEMA alleged that Suno’s models had memorized these works during training and could reproduce them on demand.

The court agreed. It found that the musical pieces were “reproducibly contained” in Suno’s models (versions 3.5 and 4), which were stored on servers in Germany. Comparing training data to generated outputs, the court concluded that the complexity and length of the reproduced pieces made coincidence impossible — the model had memorized, not merely learned patterns from, the original works.

The court treated this memorization as unauthorized reproduction under copyright law, both during training and in the outputs that users could generate with relatively simple prompts.

The Fair Use Defense Failed

The ruling is particularly significant because the Munich court applied U.S. copyright law — not just German law — to the training that occurred in the United States. Under EU private international law (the Rome II Regulation), the court determined that U.S. law governed the acts committed on U.S. soil.

Suno argued that its training constituted fair use under 17 U.S.C. § 107. The court examined all four statutory factors and found none weighed in Suno’s favor:

• Purpose and character of the use: The court found Suno did not create new, transformative musical works. Its outputs fulfilled the same purpose as the originals — listening to music. The court also noted that Suno had circumvented YouTube’s “Rolling Cipher” download protection, treating this as evidence of bad faith under the DMCA’s anti-circumvention provision.
• Nature of the copyrighted work: The outputs drew on creative elements of the original compositions, not merely facts or ideas.
• Amount used: These were not intermediate copies used in a technical process. The works became publicly accessible through Suno’s outputs.
• Market effect: The court pointed to actual market substitution, including YouTube tutorials showing users creating cover versions of well-known songs using Suno. It placed the burden on Suno to prove no market harm — a burden the court found Suno had not met.

The court distinguished the two 2025 U.S. rulings that leaned toward finding AI training to be fair use (Bartz v. Anthropic and Kadrey v. Meta). In those cases, the training data was not substantially reproduced in outputs. Suno’s outputs, by contrast, were substantially similar to the originals even from simple prompts.

What This Means for Indie Artists

If You Use AI Music Generators

The ruling does not ban AI music tools. It establishes that providers must license the music they train on. If your platform of choice has not secured licensing agreements with rights holders, it is operating in legal uncertainty — and that uncertainty can flow downstream to users who distribute or monetize AI-generated tracks.

Check your platform’s licensing status. Warner Music settled with Suno in November 2025 and entered a licensing partnership. Universal settled with Udio on per-generation royalties. Sony remains in active litigation against both. If your AI tool has no such agreements, the music it generates may be legally contested.

If You Distribute AI Music

Distributors and platforms are increasingly attentive to provenance. The GEMA ruling strengthens the position of collecting societies worldwide to pursue claims not just against AI companies but potentially against commercial users of infringing outputs. An indie artist who distributes AI-generated tracks that reproduce copyrighted elements could face takedowns, withheld royalties, or legal claims.

If You’re in the EU

The ruling arrives alongside the EU AI Act’s Article 50 transparency obligations, which became enforceable on August 2, 2026. These rules require AI providers to disclose when content is AI-generated and to publish summaries of training data used. For indie artists in the EU, this means AI music platforms operating in Europe must become more transparent about what they trained on — and collecting societies like GEMA now have both legal precedent and regulatory tools to enforce compliance.

The Jurisdictional Precedent

One of the most consequential aspects of the ruling is jurisdictional. The Munich court asserted authority over training that occurred entirely in the United States, reasoning that because Suno’s models were stored on servers in Germany and its outputs were accessible to German users, German courts had jurisdiction.

GEMA’s CEO, Dr. Tobias Holzmüller, stated: “If systems are operated in Europe, legal action can also be taken in European courts. This is crucial for the enforcement of the law.”

This means AI companies cannot escape European copyright enforcement simply by training their models outside the EU. If outputs reach EU users, EU courts may claim jurisdiction over the underlying training acts.

What Hasn’t Changed

The U.S. fair use question remains open in U.S. courts. The Sony v. Suno case in Massachusetts has been rescheduled: fact discovery runs to September 30, 2026, with dispositive motions due April 9, 2027. No U.S. fair use ruling on AI music training is expected before 2027. Earlier reports of a “summer 2026” U.S. ruling were based on misreadings of the docket.

The German ruling is also not final — Suno has indicated it may appeal. But it is immediately enforceable under German procedure, meaning Suno must comply with the injunction while any appeal proceeds.

Practical Steps

For indie artists using AI music tools:

• Audit your AI music platform. Check whether it has licensing agreements with major labels and collecting societies. If it does not, understand that tracks generated by that platform may face legal challenges.
• Document your creative contributions. If you use AI as a tool alongside your own musical input, keep records of your prompts, edits, and creative decisions. This strengthens a claim of human authorship.
• Monitor distributor policies. Platforms like DistroKid, TuneCore, and CD Baby are updating their terms around AI-generated content. Stay current on what your distributor requires.
• Be cautious with commercial release. Until the legal landscape settles — particularly the U.S. fair use question in 2027 — releasing AI-generated music commercially carries elevated legal risk, especially in the EU.
• Consider the source. Platforms that trained on licensed or public domain data are on stronger legal footing than those that scraped copyrighted material without permission.

Common Misconception

“AI-generated music is legal because the AI learned patterns, not specific songs.”

The Munich court directly rejected this argument. It found that Suno’s model did not merely learn abstract musical patterns — it memorized specific works and could reproduce them in recognizable form from simple prompts. The distinction between “learning” and “memorizing” is now legally significant. If a model can reproduce a copyrighted work when prompted, a court may find that the work was copied, not just studied.

When to Ask a Lawyer

• You are releasing AI-generated music commercially and are unsure whether your platform’s training data sourcing is legally sound.
• You received a takedown notice or claim on a track that used AI-generated elements.
• You are building or distributing an AI music tool and need to understand licensing obligations in both the U.S. and EU.
• You are an EU-based artist and want to understand how the GEMA ruling and the AI Act’s Article 50 affect your rights and obligations.
• You are a rights holder whose music may have been used to train AI models without authorization.

Sources

1. Munich Regional Court, GEMA v. Suno, Case No. 42 O 763/25, Judgment of July 31, 2026 — Summary via kpw.law
2. GEMA press release, July 31, 2026 — Via Variety
3. Detailed legal analysis of the 143-page judgment — SKW Schwarz
4. Chartlex Music Industry AI Lawsuits Tracker, verified August 20, 2026 — chartlex.com
5. EU AI Act Article 50, transparency obligations effective August 2, 2026 — artificialintelligenceact.eu
6. EU Commission Guidelines on AI Transparency Obligations — digital-strategy.ec.europa.eu