German collecting society GEMA has released a copyright-cleared music dataset for training artificial intelligence models, offering developers a single source of licensed audio files, metadata, and rights.
The initiative, called PLAI by GEMA, bundles sound recordings with both composition and master recording rights to simplify licensing and reduce legal risks for AI companies. GEMA describes the dataset as the first of its kind in Europe.
The launch arrives as GEMA awaits a ruling in its copyright lawsuit against AI music generator Suno, which the society alleges trained its tool on unlicensed copyrighted recordings.
AI music transcription company Klangio is the first licensee. The dataset provided to Klangio comprises approximately 178,000 recordings spanning 60 genres.
Thomas Theune, GEMA’s Director of Broadcasting/Online, said:
“With PLAI by GEMA, we are making music content for training AI tools more easily accessible than ever before. Instead of complex rights clearances, AI tool providers now receive high-quality music data, including all necessary rights, from a single source. This creates legal certainty for developers and new revenue potential for music creators.”
“With PLAI by GEMA, we are demonstrating that AI training and fair compensation for creative professionals are not mutually exclusive. If both sides are interested in responsible collaboration, we can find common solutions and enable innovation on a reliable legal foundation.”
