On 31 July 2026, the Munich Regional Court ruled against Suno in the case brought by GEMA, the German collecting society for composers, lyricists, and music publishers. I have not obtained the written judgment, so the account below relies on the court’s official release. According to the release, the case concerned six musical works, and lyrics were not at issue.
It has been reported as a win for musicians. Performers’ rights in the recordings were not among the claims described in the court’s release. That matters, but not for the reason usually given.
A client can replace a session guitarist without the generated track copying any performance anyone can point to. That is the gap. Nothing in this ruling closes it.
The player is not legally invisible
The guitarist has a legal category, and it is worth disposing of the opposite claim before making the argument that survives.
German law recognises the performer directly. Under sections 74 to 78 of the Copyright Act, a performer has rights of attribution and integrity, and rights in the fixation of a performance, its reproduction and distribution, its making available, and its broadcasting and other public communication. Those rights attach to the performance, not merely to one master considered as an object, and they sit separately from the phonogram producer’s rights under section 85. GVL administers performers’ secondary exploitation rights and statutory remuneration collectively, while primary exploitation is generally governed by contracts between rights holders and users.
What the guitarist lacks is narrower than that, and harder to name.
Where the gap sits
Existing rights are built around identifiable things. A composition. A particular fixed performance. A name, a likeness, a voice, an implied endorsement. Each of those can be pointed at.
A session guitarist sells the ability to produce the right feel, tone, and phrasing on demand. That ability is built over decades from small, repeatable judgements, including a particular attack on the string, a habit of landing slightly behind the beat, and an instinct for what the track requires. Taken together, those qualities have commercial value, but they are not a composition, a fixed performance, or a name, and those are the things the law knows how to point at.
In twenty-five years around studios and instruments, I have seen guitarists booked not because a client wanted one particular take repeated, but because they could hear what the track needed, find the part quickly, and deliver it before the clock became expensive. The same is true of drummers, bassplayers, horn players etc., and I write about guitarists because that is the working economy I know.
The mismatch between what a client buys and what copyright protects is where the enforcement gap appears. A generated track might copy no composition, sample no master, reproduce no substantial part of any single take, and invoke no player’s name, and still be bought instead of the session. In that transaction, copyright may find no protected work that has been copied. The client still finds a substitute.
Whether a model isolates one guitarist’s playing or approximates it through correlations drawn across genre, production, and many different performers is technically unresolved, and the Munich court did not address it. The output only has to be good enough to take the job.
What other rights can do
Existing law may help where the imitation uses a player’s identity, suggests endorsement, breaches a contract, or copies an identifiable performance. American courts recognised claims over deliberate imitation of a distinctive voice in advertising decades before generative models, in the Midler and Waits cases, and several American states have since enacted digital replica laws reaching voice and AI impersonation. In Germany, personality rights and unfair competition law may become relevant where an imitation is falsely attributed to a known player.
None of those routes gives a guitarist ownership of a way of playing. They run on identity, deception, and endorsement, and each requires proof that something recognisably theirs was taken. Voice legislation shows that lawmakers can protect a recognisable personal attribute when it is tightly bound to identity. It also shows why the instrumental case is harder, since a voice is far easier to identify evidentially than a composite formed jointly by a player, a guitar, an amplifier, a room, a producer, and a song.
Sound-alike advertising, tribute recordings, sample libraries, and session substitution have generated disputes for decades. Scale, cost, and the speed at which a buyer can obtain a substitute have all changed.
Where this lands in the trade
The exposure falls unevenly, and I do not think it falls where the trade press assumes.
The category I expect to be most exposed is the middle tier of commercial guitar work, meaning advertising, library music, demo work, and production cues, where the commission is priced on convenience and the performer’s identity is usually not part of what the client is buying. That is the work that has historically funded serious players between the visible jobs. I am inferring that from the economics of convenience-priced work, without employment data behind it, though the competitive logic is clear.
Physical instruments sit outside this argument. Custody, verified manufacture, material connection to a period, finite supply, and the experience of playing the thing are not things a model reproduces, and Vintage & Rare has no evidence, as of 2026, that generative audio has moved the vintage market in either direction. What is exposed is the income of the people who made those instruments matter, which is a different question from what the instruments are worth.
Nothing is settled
The judgment is not final, Suno has said it is evaluating all options including an appeal, and related European litigation remains unresolved. None of it currently supplies a clear right in the recurring characteristics of a guitarist’s playing.
Organised rights holders and larger commercial claimants are better equipped to litigate these disputes, with registration systems, distribution mechanisms, and the capacity to sustain a lengthy case. Any eventual licensing settlement will reflect who was able to enter the room. Individual guitarists are among those least equipped to do so, and what they would ask the law to protect remains unusually difficult to define.
The law can protect the performance. It may protect the player’s identity. The accumulated judgement a client is actually paying for falls between them.
Written by Nico von Cotta, Vintage & Rare
Nico von Cotta founded Vintage & Rare in 2008 and has spent more than 25 years making music as a guitarist, singer, songwriter, and producer.
What the court decided, in more detail
According to the official release, GEMA alleged that reproductions occurred during training in the United States and that the works were memorised within the model itself. The court assessed the American training activity under American law and rejected Suno’s fair use defence on the facts before it, which is a narrower proposition than the idea that European copyright follows a developer wherever it trains. It treated memorisation within a model hosted on German servers as a separate matter under German law. It attributed the outputs to Suno rather than to the users because the prompts did not specify melody, harmony, rhythm, or arrangement, while Suno operated the models, controlled their architecture, and was responsible for the memorisation of the training data. The claims were largely granted. The release gives no indication that performer or recording rights were pleaded, though the absence of a mention in a summary is not the same as verified confirmation that no such claim appeared anywhere in the pleadings.
Evidential limits
This article is written from the Munich Regional Court’s official release of 31 July 2026 in case no. 42 O 763/25, together with GEMA’s statement, reporting by Variety, Music Week, Music Ally, and Reuters, and published legal commentary. The German performer rights summary is drawn from sections 74 to 78 and section 85 of the Copyright Act. No verified copy of the written judgment or of the English translation reported to be circulating has been obtained. Nothing here should be read as a description of the court’s full reasoning.