A first-of-its-kind licence could give creators control and compensation when their work is used by AI.
Swedish collecting society STIM has taken a major step in how music and artificial intelligence intersect. Last week, it introduced what it calls “the world’s first collective AI license for music,” designed to ensure creators are fairly compensated when their work is used to train AI models.
The licence applies only to works from creators who have explicitly consented to participate. STIM explains that it “ensures fair compensation when their music is used in AI contexts,” giving artists control over how their music contributes to AI development. With concerns that up to 24 percent of music creators’ revenues could be affected by AI developments by 2028, the licence provides a structured path for creators to protect their earnings.
As Music Ally reports, this rollout is currently a pilot project with startups Songfox and Sureel. STIM notes that “only a limited number of works” are included in this first phase. A standout feature is the use of independent attribution technology, which allows AI-generated music to be traced back to the human-created works that influenced it. This ensures royalties are distributed accurately and transparently throughout the AI training process, as well as when the generated music is used commercially.
Lina Heyman, Interim CEO at STIM, said: “We’ve created a license model that enables us to respond to AI developments without compromising the rights of music creators. It’s about ensuring that anyone who wants to contribute their music to AI contexts is fairly compensated and has full transparency. The license becomes a framework for how copyright and innovation can go hand in hand – with fair terms for creators and clear rules for AI companies.”
STIM’s initiative builds on a wider industry discussion about licensing and AI. In October 2023, French society Sacem opted out of allowing its members’ works to be used for AI training to encourage proper licensing. Music Ally notes that similarly, German society GEMA has emphasised that “Pure remuneration through a buyout, i.e. a one-off lump sum payment for training data, is not nearly sufficient to compensate authors in view of the revenues that can be generated.” STIM’s approach mirrors this philosophy.
Discussions around AI and fair pay are growing across the industry. As the debate over whether (and how) artists should get paid for AI usage continues, understanding the ethical and financial implications of AI using music is becoming increasingly important. At the same time, ongoing negotiations between AI startups and record labels show how licensing frameworks, like those explored with Suno and Udio, are evolving to give creators proper protection.
While launching the licence is a decent step forward, its success depends on AI companies actually adopting it. STIM’s framework offers a way to embrace new technology without compromising creators’ rights, providing transparency and fair compensation at every stage. By establishing clear rules for using music in AI, it creates a path for innovation that respects the work of musicians while giving AI developers a legal and ethical way to train their models.