GEMA, Germany’s performing rights organization (PRO), has filed a copyright infringement lawsuit against OpenAI in Munich’s regional court. GEMA alleges that OpenAI used song lyrics it licenses without obtaining proper permission. Notably, the case centers solely on lyrics, excluding any song recordings.

GEMA calls the lawsuit a “model action,” which they hope will help define copyright laws not only in Germany but potentially across Europe. The organization claims that OpenAI’s software, in responding to user prompts with song lyrics, may have been trained using lyrics for which GEMA controls licensing rights.

GEMA argues that OpenAI has infringed on two main rights: the “making available” right (a right under European law akin to public display or performance in the U.S.) and reproduction rights. According to GEMA, the infringement occurs in two instances—first, when lyrics are ingested for AI training and again when they’re reproduced as responses.

Unlike in Germany, U.S.-based PROs don’t control mechanical rights, so they would not be able to file a similar lawsuit.

Billboard explains that most music industry lawsuits against AI companies have centered around using recordings, though that process also inherently includes the underlying compositions. Meanwhile, OpenAI is dealing with its own legal battles on multiple fronts. It’s facing a potential class-action suit from authors, legal action from The New York Times and separate lawsuits from online publishers and other news outlets.

In Europe, copyright law is framed around “exceptions and limitations” rather than a broad concept like fair use. Under the 2019 Copyright Directive, text and data mining are generally allowed—unless copyright holders choose to opt-out.