Harry, sing Lana Del Rey! How AI is making pop fans’ fantasies come true

Last month, a video of Harry Styles covering a song from 2003 Disney film The Lizzie McGuire Movie went viral on TikTok. Except Styles has never – at least not publicly, or to anyone’s knowledge – performed this song. Instead, it joins the many examples of fake performances created by artificial intelligence.

Fans were delighted. “Suddenly I am no longer afraid of AI,” commented one listener, and on videos of other AI-generated Styles songs (he’s made to duet with Taylor Swift, for example) there are multiple requests for more. “Make him sing take me to church [by Hozier],” writes one fan, and “SING SAD GIRL BY LANA [del Rey]”, bellows another. Fans wanting their favourite artists to cover songs they enjoy is hardly a new phenomenon, but with advances in AI, there’s no need to involve the artist at all.

Musicians are therefore worried – about being made to perform material they otherwise wouldn’t, or being usurped by a fantasy. “I can’t help but think that I can be easily replaced,” says Flora Rose, a singer-songwriter on TikTok. “I’m spending months crafting my debut EP, [and meanwhile] people can make tracks in one click.”

When it comes to the arts, AI tends to provoke horror or ridicule – as when an AI photograph won a major photography competition, or when ChatGPT declared young adult weepie The Fault in Our Stars “one of the best books of all time”. In February, the lawyer behind a lawsuit on behalf of visual artists whose work was being used to generate AI art called any generative image “an infringing derivative work”.

The music industry has taken a similar stance. Last month, Universal Music Group asked streaming platforms to decide whether they wanted to be on “the side of artists, fans and human creative expression,” or “the side of deep fakes, fraud and denying artists their due compensation”.

The problem with UMG’s statement is that it assumes fans and artists are on the same side of this debate. A report by JP Morgan in April argued that “AI music is just not very good,” and “people don’t listen to it.” But the popularity of AI covers – over 9m views on a fake collaboration between Drake and the Weeknd before it was taken down – puts this into question, and in the comments of these covers on TikTok, sentiment is overwhelmingly positive. Phoebe, 23, a Styles fan from London, says she loves “the idea of hearing Harry cover one of my favourite songs, without relying on him to sing it on stage”.

Singer-songwriter Victoria Canal says she doesn’t think AI covers can “replace the heart of an artist’s original voice. It’s fun and shocking to hear an AI Drake song, but I don’t think it has the potential to take work away from Drake”. But AI music is still sounding alarm bells across the industry: Drake recently called an AI cover of himself rapping an Ice Spice song “the last straw”.

That JP Morgan report insists that “superfans fall in love with the artist, not just the music.” Canal agrees, saying AI “can’t replace the heart of an artist’s original voice and recording, and people can feel that”. But existing artists with strong personal brands such as Styles, Drake, or Taylor Swift, already have an emotional connection in place with their fans. “Pop stars are a projection of our desires,” says Holly Herndon, a musician who has experimented with Holly+, an AI clone of her own voice. “We resonate most with idealised fantasies”, she says, rather than the artist themselves.

AI covers are “linked to [Styles], but separate,” Phoebe says. “Being a fan of someone is about the community around it, and I love how AI covers give us more content to talk about.”

But it’s not just fan service which AI is facilitating. Stan culture – fervent internet-based fandom – has become increasingly critical and demanding of its idols. Consider the online reaction to Frank Ocean, whose radio silence over new music annoyed fans long before his Coachella set was found wanting, or to Charli XCX, who is often besieged by fan criticism of new music or concert setlists.

Fandoms often “treat pop stardom like sports,” says DJ Louie XIV, host of the pop-music podcast Pop Pantheon. “You pick a favourite artist and become very invested in the idea of them ‘winning’ over the others in the field.” As fans ask for more, sometimes at the cost of an artist’s own creative freedom, AI could be satisfying those needs more than the artists themselves.

But many fans see AI covers as an opportunity for collaboration and communication. “AI songs could give artists the chance to grasp what their fan bases want from them,” says Sara, who runs a Taylor Swift fan account on Twitter. Herndon believes that in the near future, it will be just as common for artists to “assume the identities of other people” as it is to use samples, and “artists will experiment with letting others perform as them, and share in the profits”. UMG and Drake might not agree – but last week, Canadian singer Grimes tweeted that she would split royalties by 50% on any successful AI song using her voice. “I like the idea of open sourcing all art and killing copyright,” she wrote.

But as AI programs advance, it will become difficult to tell the difference between real and generated music – and fans could end up horrified by what their idols are being made to sing. Grimes later wavered, writing in another tweet a few days later that she “may do copyright takedowns … don’t wanna be responsible for a Nazi anthem.”