Spotify boss says AI music outperforms poor-quality tracks

Spotify’s chief executive, Alex Norström, has defended the streaming giant’s push into AI-generated music, arguing that its controlled system offers a better alternative to piracy and the “unregulated AI slop” that has already begun flooding online platforms. Speaking to the Financial Times, Norström said the company was trying to provide a framework in which musicians can consent to the use of their work and earn money from it, rather than seeing it pirated or used without permission. “There’s a lot of rogue attempts at this,” he warned.

The new feature, announced last week as part of a deal with Universal Music Group (UMG) that sent Spotify’s shares up 16%, will allow premium subscribers to create their own AI-powered remixes and song covers using music from participating artists. Norström described the initiative as a way to enable “one song to become 10,000”. The feature will not be included in a standard Spotify Premium subscription; instead it will be offered as a paid add-on, though the exact price has not yet been disclosed. Participating artists and songwriters will share in the revenue generated, creating an additional income stream on top of existing royalties.

Both Spotify and UMG have emphasised that the initiative is “grounded in consent, credit, and compensation”. Artists can choose to opt in, and those who do not will effectively block access to their work for these AI creations. However, several operational details remain unclear. It is not known whether user-generated AI remixes will be shareable with other listeners or kept private, and it is also uncertain how Spotify might label such content if it is made public. The company has introduced a voluntary AI tagging system, currently in beta, to indicate AI involvement in music, but experts note that voluntary disclosure can be problematic if dishonest users avoid tagging, and there is no verification mechanism in place.

Industry concerns over competition and displacement

Ed Newton-Rex, a composer and campaigner for protecting artists’ copyright, welcomed the focus on consent but warned that the feature could — depending on its design — create a “vicious circle” in which human artists face greater competition from AI-generated work. “I think if you are going to have AI music, it’s clearly better that you have AI music that is rooted in consent,” he said. “The big question will be whether fans can share remixes they make for other people to listen to. If they can, I think you get into dangerous territory. These AI remixes will flood Spotify and drown out other songs, which will in turn put pressure on more musicians to sign up to the AI remix feature.”

The potential for AI music to compete directly with human artists is a central concern. Newton-Rex pointed out that there are only so many hours in a day for listening to music, meaning every listen to an AI-generated track is a listen not given to a human artist. “The framing is absolutely AI music versus human music,” he said. “Whenever someone listens to AI music on Spotify, they are not listening to a song that is simply made by a human.” This tension is underscored by clear market demand: last year three AI-generated songs topped music charts, including on Spotify, highlighting how AI music has become increasingly indistinguishable from human-made work. “AI music has gotten much better very quickly, and so if people hear a song they like and they’re not thinking about whether or not it’s AI, then there’s demand for it like there’s demand for any music,” Newton-Rex added.

Some Spotify users have already expressed fears about the platform’s growing reliance on AI, with a number threatening to quit the service. Critics argue that AI tools diminish the value of human creativity and the effort involved in making music. The broader legal landscape is also fraught. In the UK, the government backtracked in March on a plan to let AI companies use copyrighted work without permission after a backlash from thousands of artists, including Elton John and Dua Lipa. Leading AI companies such as Meta and OpenAI are facing multiple lawsuits from authors, publishers and artists over the alleged use of books, newspaper articles and other copyrighted materials to train AI models without consent. In a new development last week, two Meta employees were individually sued for allegedly pirating a terabyte of books to train the Llama model.

Newton-Rex argued that Norström’s attempt to frame Spotify’s move as a choice between curated AI content and unregulated AI slop sidesteps the real issue. “The framing is absolutely AI music versus human music,” he repeated, stressing that the most immediate competition is not between legitimate and rogue AI, but between AI and human artistry. The Spotify-UMG deal, which mirrors similar “walled garden” arrangements UMG has struck with platforms like Udio and Warner Music, is designed to keep AI-generated content within licensed, controlled environments — but whether that will protect human artists or accelerate their displacement remains an open question.