Six Songs at the Center of the GEMA–Suno Decision

How Our Research Connects to the GEMA Case

‍

The GEMA case centered on two questions:

1. Were protected songs used to train Suno’s AI?
2. Did recognizable parts of those songs appear in its generated music?

‍

Sound Ethics investigates both sides of the issue, the music used to train AI models and the material those models reproduce. Our research platform helps identify specific songs used to train models and advance AI and machine-learning research. Together, Sound Ethics and research teams at NYU, UC Riverside, Cal Poly, and UC Berkeley compare human works with AI-generated music, uncover songs used in AI development and test which methods for this work best.

‍
‍

How GEMA Built Its Evidence

GEMA's evidence was practical: repeated song-targeted prompting. The central question was whether generated music had become too close to known songs.

‍

Why This Research Matters Now

Sound Ethics' research in this area provides critical support at a moment when artists, rights holders, developers, and courts need clear answers to these questions.