Should the Dead Get an Encore?
Subhanga Khanal | August 2026
Here is a moment I cannot stop thinking about: I clicked on an AI song of Freddie Mercury, the late frontman of Queen, singing a track he never recorded, and I absolutely loved it. Then I closed the tab wondering if he would have loved it too.
AI can now listen to old recordings of a singer, learn exactly what their voices sound like, and then make that voice say or sing things it never did. Think of it as a musical impressionist that never gets the voice wrong. For living performers, this raises familiar questions about consent and ownership. For performers who have died, however, the problem is more complicated. Artists who passed away years ago can, technically, still release new material. Yet they can no longer consent to this new performance, but their heirs and rights-holders may still have legal and commercial interests in their name, image and work.
The Beatles’ ‘Now and Then’ offers one of the more accepted versions of posthumous technology. The song began with a home recording made by John Lennon and was completed by the surviving Beatles decades after his death. AI-assisted audio restoration helped separate Lennon’s voice from the original recording so, it could be incorporated into the finished track. The result was celebrated by many fans, although the project also raised a question that will become harder to avoid as synthetic voices become more capable: where does restoring an artist’s original performance end and creating a new one begin?
The case when Warner Music Group and the estate of Edith Piaf agreed to use AI to recreate Piaf’s voice and image for an animated film about her life draws a clearer line. The project uses her reconstructed voice for narration and was developed with the involvement of her estate. The purpose is not to manufacture a new Piaf recording and present it as an undiscovered performance, but to tell her story using a digitally reconstructed version of the person audiences already know.
The case of the Notorious B.I.G., while similar, is more complicated. In 2023, producer Timbaland shared a track using a generative AI recreation of the rapper’s voice. He described it as an opportunity to collaborate with an artist he had never been able to work with during his lifetime. The experiment generated enthusiasm but also revived the question of whether a deceased artist’s voice can be used to create something the artist never performed or approved.
The distinction between these cases is not really technology. It is instead authority.
An estate may have the legal right to license aspects of an artist’s identity, but that does not necessarily mean that it can answer every question about what the artist would have wanted. An artist might have left explicit instructions about posthumous use of their voice. Yet another might have left nothing. A third might have wanted their recordings preserved but objected to their voice being used to perform entirely new material. The use of AI in the recreation of voices and identities makes these distinctions commercially relevant because the difference between an old recording and a synthetic new performance is no longer necessarily audible.
The law, in some countries, is beginning to respond to this problem. Tennessee’s 2024 Ensuring Likeness, Voice and Image Security Act, known as the ELVIS Act, expanded the state’s existing publicity-rights framework to include a person’s voice. The law allows the right to be exercised after death by executors or administrators, heirs, assigns or devisees, generally for 10 years after death and subject to continued commercial use and establishes civil liability for certain unauthorized uses of a person’s voice, including some uses involving AI-generated replicas. It also provides exceptions for protected forms of expression such as commentary, criticism, scholarship, satire, and parody.
The significance of the law goes beyond music. It reflects a broader shift in how policymakers are beginning to think about identity in an age when a person’s appearance, voice, and mannerisms can be reproduced without their physical presence. Copyright law protects particular works. Personality and publicity rights protect aspects of the person. AI increasingly blurs the boundary between the two.
There is no obvious reason to prohibit every posthumous use of AI. Reconstructing a voice for a documentary, completing an unfinished recording with the artist’s prior approval, or creating an authorized biographical work is different from putting words into a dead performer’s mouth for commercial purposes without permission. Treating all these cases in the same way would be as unhelpful as treating them as entirely unrelated. The more useful policy question is therefore not whether the dead should be allowed an encore. It is who gets to decide what that encore sounds like.
As synthetic media becomes cheaper and more convincing, artists may need the ability to specify how their voices and likenesses can be used after death. Estates and heirs may need clearly defined rights and responsibilities. Audiences, meanwhile, may need to know when a performance they hear was generated rather than recorded by the artist themselves. AI can now make a dead artist sound alive. The harder question is how much authority over that voice should survive the artist.
Subhanga Khanal is a Jr. Research Assistant at IIDS.
The views and opinions expressed in these articles—authored by IIDS interns and staff—are solely those of the individual writers and do not necessarily reflect the official policy, position, or endorsement of the Institute for Integrated Development Studies (IIDS).
