[Re: Google and Universal in talks to license music to be use by AI,
Generative AI raises all kinds of questions about the adequacy of copyright law in the face of new technological capabilities. What measures can really be implemented to stop AI algorithms from being trained on existing music? When Google recently published its first text-to-music algorithm, it was said to have been trained on 240,000 hours of recordings. But the majority of the rights to reproduce that music would be held by music labels and publishers. So while the negotiations currently underway might be about the right to publish something that purports to be Drakelike, the music industry will be keen to assert its right to licence the act of AI training in the first place.
Encouragingly, some musicians like Grimes are exploring collaborations, actively licensing their voiceprints for AI apps, and encouraging creatives around the world to generate music using their voice and sound. These choices become ethical as well as creative.
Jeremy Silver Digital Catapult
(c) 2023 City A.M., source