AI Voice Clones in Music: Who Gets to Use an Artist’s Voice?
AI voice cloning can recreate a singer without them entering a studio, but artists, platforms and lawmakers are increasingly drawing a line around consent.
Artificial intelligence can now reproduce a singer’s voice closely enough for new performances to be created without the artist entering a recording studio. For the music industry, that has turned a technical development into a much bigger question about consent, identity and who should control a recognisable voice.
Some musicians are choosing to license AI versions of themselves, while streaming and video platforms are introducing protections against unauthorised copies. The difference between those two situations is becoming one of the clearest dividing lines in the debate around AI music.
Grimes showed what an authorised AI voice could look like
Grimes became one of the earliest major artists to openly experiment with licensing her voice when she launched Elf.Tech in 2023. The system allowed creators to upload or record vocals and transform them into a voice resembling hers.
Creators could commercially release approved results under an arrangement requiring a 50 per cent share of master recording royalties for Grimes. It offered an early example of AI voice technology being used with the knowledge and participation of the person whose identity was being reproduced.
That matters because the same technology creates very different concerns when the artist has not agreed to take part. A licensed digital voice and an unauthorised imitation may sound similar to a listener, but the consent behind them is completely different.
Spotify now restricts unauthorised vocal impersonation
Spotify strengthened its impersonation rules in 2025 and says uploaded music containing an AI voice clone of another artist is only permitted when that artist has authorised the use. The policy also applies to other forms of unauthorised vocal impersonation rather than AI alone.
Spotify says artists can report releases that clone their voice without permission, including cases where the uploader presents themselves as an AI version of that performer. The platform can remove content when it finds that its impersonation policy has been breached.
The rules are separate from Spotify’s newer AI Persona system, which focuses on whether an artist profile itself represents an artificial identity. A real singer whose voice is copied without permission presents a different issue from a fictional performer that openly identifies itself as digital.
YouTube is developing similar protections
YouTube has explored authorised AI music through projects including Dream Track, which allowed selected creators to generate short pieces of music using AI versions of participating artists’ voices. The original experiment was developed with Google DeepMind and involved artists who had agreed to take part.
At the same time, YouTube has expanded its likeness detection technology to the entertainment industry. The system helps eligible entertainers, agencies and management companies find AI generated content using a person’s likeness and request removal where appropriate.
Together, these approaches show two very different uses of the same technology. Artists can choose to participate in authorised AI experiments, while platforms are also building tools to help them challenge uses they did not approve.
Why voices raise different concerns
Using AI to create a synth sound or assist with mixing does not necessarily involve another person’s identity. Cloning a recognisable singer raises a more personal issue because it can make it appear that somebody performed words or music they never actually recorded.
That can affect reputation as well as income, particularly if a synthetic performance places an artist’s voice into material they would never have agreed to support. The legal position can also vary depending on the country or US state because voice, likeness, copyright and publicity rights do not all operate in the same way.
US lawmakers are considering new protections
The NO FAKES Act of 2026 is one attempt to create clearer federal protection in the United States for unauthorised digital replicas of people’s voices and visual likenesses. The bill was introduced in May 2026 and was approved by the Senate Judiciary Committee in June before being placed on the Senate legislative calendar.
It has not become law, and further action would still be required before it could take effect. The proposal includes civil protections against certain unauthorised digital replicas while also providing exceptions for uses including news reporting, sports broadcasts and satire.
Consent may become the clearest dividing line
AI voice cloning is unlikely to disappear, and authorised versions could create new ways for musicians to collaborate with fans and other creators. An artist could potentially license a digital voice for projects they choose to support while retaining control over the circumstances in which it is used.
The concern is what happens when that choice disappears. As platforms develop more powerful AI tools, the most important question may not be whether a voice is artificial, but whether the person it belongs to knew about the use, agreed to it and shares in the value it creates.
Visit SoapLandTV’s AI & Entertainment section for more Industry & Debate features, Media stories and Entertainment News exploring how creators, platforms and audiences are responding to the technology.
