GEMA Sues AI Audio Generator Suno Over Copyright Infringement

GEMA : GEMA Sues AI Audio Generator Suno Over Copyright Infringement

The Society for Musical Performing Rights (GEMA) has filed a lawsuit against the AI audio generator Suno at the Munich Regional Court. GEMA accuses Suno of unlawfully training its generator using copyrighted music works. With this lawsuit, GEMA aims to ensure that companies like Suno pay license fees to the creators for using music. According to GEMA, songs such as “Forever Young” by the 80s synth-pop band Alphaville, “Atemlos” by Kristina Bach, the Boney M. song “Daddy Cool” written by pop producer Frank Farian, and works by the Modern Talking duo Dieter Bohlen and Thomas Anders are specifically affected. GEMA believes these and other titles are being replicated in melody, harmony, and rhythm.

“AI providers like Suno Inc. use the works of our members without their consent and profit financially from it,” says Tobias Holzmüller, Managing Director of the society. “At the same time, the output generated competes with human-created works and undermines their economic foundation.” Suno, based in Cambridge, Massachusetts, currently charges its users $10 per month for a pro version. Users receive credits to generate audio through prompt input.

GEMA seeks to create legal clarity with this lawsuit, increasing pressure on generative AI service providers. Due to the app’s availability throughout Germany, jurisdiction is freely selectable. GEMA itself is based in Munich and had already filed a lawsuit against OpenAI there last year. However, the accusation was not that the ChatGPT developer used the music unlawfully, but that copyright was violated in the texts, as no licenses were acquired for them. The society, which has 95,000 members, aims to enforce licensing by AI companies through these proceedings.

GEMA is not completely sure that their approach will succeed in court: “If we do not want to forego human-made music in the future, we urgently need a legal framework that ensures creators a fair share in the value creation by AI providers,” says Supervisory Board Chairman Ralf Weigand. The uncertainty stems from the question of how far an exception rule written into European law and German copyright law in 2019 for training AI using text and data mining applies to such cases. This affects all types of copyrighted works. Societies are currently searching for answers. The discussion about whether and to what extent AI can be trained with legally protected material is also in full swing in the USA.