Google-owned YouTube is planning to continue its discussions with major record labels to expand its AI music creation capabilities. The platform’s Dream Track project, launched last year with the support of 10 artists, demonstrated the potential of AI to create original music. Instead of expanding on this project, YouTube is now seeking permission from major labels to clone even more musicians.
YouTube aims to license songs from “dozens” of artists to train new AI tools that will be launched later this year. In exchange for licensing songs for training purposes, YouTube is prepared to pay substantial sums of money to Universal Music Group, Sony Music Entertainment, and Warner Records. However, the exact amount of the licensing fees has not been disclosed yet.
The payment strategy adopted by YouTube may be aimed at overcoming potential resistance from record labels. Despite YouTube’s eagerness, both artists and labels may need time to be convinced. For instance, Sony Music has cautioned against unauthorized use of its content by AI companies in the past. UMG also temporarily removed its music catalog from TikTok due to concerns surrounding AI-generated music protection. In a contrasting move, more than 200 artists have called on tech companies to halt the use of AI to infringe upon their rights as human artists.
YouTube’s negotiations with major labels come shortly after the Recording Industry Association of America (RIAA) filed copyright infringement lawsuits against two prominent generative AI music companies: Suno and Udio are accused of using unauthorized copies of sound recordings to produce music on a large scale. The labels are seeking damages of up to $150,000 per infringement in these lawsuits.
The delicate balance between technological innovation and protecting artists’ rights is evident in this ongoing debate. As YouTube continues its quest for new ways to create original music using AI technology, it must navigate through legal challenges while respecting the rights and concerns of both artists and record labels alike.
