AI, voice cloning, and the future of music royalties - CMS.law

In January 2026, a legal dispute emerged concerning the viral track I Run, which allegedly utilized AI to manipulate a producer's vocals into a female tone reminiscent of British singer Jorja Smith. The case highlights the escalating tension between AI-assisted creativity and intellectual property rights, specifically regarding the unauthorized use of an artist's signature sound.

What happened

The producer, Haven, created the track by using AI prompts to emulate soulful vocal samples, claiming the original vocals were their own. However, Smith’s label, the Orchard, asserted that the AI-generated output was an imitation of the singer's highly recognizable voice, potentially misleading the public into believing she performed or endorsed the song.

Following the track's success—which included reaching #11 on Spotify US and gaining significant traction on TikTok and Instagram—the Orchard, alongside the RIAA and IFPI, issued takedown notices, resulting in the song's removal from streaming platforms. A central point of contention is whether the AI model used to create the track, identified as Suno, was trained on Smith’s copyrighted works without authorization. Haven has argued that if the model was trained on her music, they should be entitled to a share of the royalties, though the legal liability for such claims remains complex. The dispute underscores the difficulty of distinguishing between artistic inspiration and copyright infringement, as well as the challenges of applying existing intellectual property frameworks to AI-generated content. Questions persist regarding whether imitating tone, phrasing, or expression constitutes passing off or moral rights violations, especially when content lacks clear AI labeling to inform the listener.