Sony Group is introducing proprietary attribution technology designed to reverse-engineer the “underlying music” within generative AI outputs. The system attempts to quantify specific training data contributions—hypothetically attributing weighted percentages to original works—to determine exactly how much a specific catalogue influenced a synthetic track. While the reported examples are simplified, the technology aims to parse the complex, multi-input nature of latent space generation to identify source material.
This development represents a shift from passive detection to active, baked-in attribution, with Sony envisioning AI companies integrating this tool directly into their model architecture. By establishing the metrics for contribution early, Sony is positioning its technology as the potential standard for the inevitable licensing negotiations and royalty settlement layers between rights holders and generative platforms.
Curated by MusicResearch.com from Music Ally. Read the full article at: Sony tech aims to identify ‘underlying music’ in GenAI tracks
