Generative AI Music Attribution Rethinks Royalties

- Streaming royalties are historically split between recording owners (labels/artists) and composition owners (songwriters/publishers), but generative AI now creates music that mimics specific artists' styles, raising the question of who should be paid—and how much. - Current systems like Shazam and SoundCloud's catalog matching can identify direct samples or covers, but they struggle when AI output is "in the style of" a known artist without copying a specific recording. - The article likely discusses proposed technical frameworks (e.g., watermarking, fingerprinting, or attribution models) that would let AI-generated tracks credit and compensate the human artists whose works influenced the training data, similar to how sampling royalties work. - This sits at the intersection of copyright law, music industry economics, and AI policy—key debates include whether training on copyrighted data is fair use, and whether attribution should be mandatory or voluntary.