News · industry — By WATMM Desk —

Sony has asked for 260,000 AI deepfakes to be taken down, and indie artists have no Sony

Sony Music has now asked platforms to remove more than 260,000 AI tracks imitating its artists, up from 135,000 at the end of March. That is roughly 20,000 requests a month, which is a full-time enforcement operation. An independent artist facing the same impersonation has a report form, a distributor and their own spare time.

Key facts

  • What happened: by the end of September 2026 Sony Music had asked platforms to remove more than 260,000 AI-generated tracks imitating its artists, according to the Financial Times, nearly double the 135,000 it reported at the end of March.
  • How: Music Business Worldwide works this out at more than 20,000 requests a month across April to September. Sony has not said how many tracks were actually removed or how fast, and its count does not separate voice clones from music delivered to the wrong artist profile.
  • Why it matters: an independent artist gets the same problem without the staff. On Spotify, a voice-impersonation claim has to come from the artist or their representative, and the tool that would catch fakes before release now lets them go live unless the artist objects.

Jason Moran found out about "his" new EP from a friend. The jazz pianist told Digital Music News in April that the release on his Spotify profile had no piano on it at all. He reported it, and Spotify took it down about 72 hours later. His complaint was not the speed. It was that the fake sat there until he happened to hear about it, and that the job of finding it fell on him.

Hold that next to this week's number. By the end of September, Sony Music had asked platforms to remove more than 260,000 AI tracks imitating its artists, according to the Financial Times. The artists named include Harry Styles, Adele, Britney Spears, Queen and Michael Jackson. Sony's total was 135,000 at the end of March, so it has nearly doubled in six months.

What the number does and doesn't say

Music Business Worldwide puts the pace at more than 20,000 takedown requests a month across April to September. The earlier figures show how fast that has grown. Sony said it had sent close to 10,000 takedowns by November 2023 and more than 75,000 by March 2025, and the BBC reported that about 60,000 of the 135,000 were flagged after March 2025. That works out to roughly 5,000 a month over that year. If those figures are comparable, the monthly rate has roughly quadrupled. That is our arithmetic, not Sony's.

Three things are missing, and MBW points them out:

  • Outcomes. The FT didn't report how many tracks were actually removed, how quickly platforms acted, or which services got the requests. A request isn't a removal.
  • What counts. Sony's tallies don't separate unauthorised AI voice clones from music fraudulently delivered to an artist's profile. Those are different problems with different fixes.
  • The platforms' side. The FT report carried no comment from streaming services.

Sony told the FT that the tracks imitate its artists' voices and likenesses without permission and that it expects platforms to act quickly. It also called the work "an uphill struggle". Sony's Dennis Kooker told Music Week in May that deepfakes are "demand-driven": they turn up when an artist is promoting a release and feed off the attention the artist built. In other words, the fakes arrive during release week, when an artist has the least time to spare.

What a major can do

The clearest example is "I Run" by Haven, whose AI-processed vocals were compared to Jorja Smith. MBW, citing Billboard, reports that Sony-owned distributor The Orchard sent takedown notices, and so did the RIAA and IFPI. Smith's label, FAMM, said the track had gone out through four different distributors, which it called an attempt to get around the usual takedown process. Spotify said it removed the track and paid no royalties on its streams. A major can bring a distributor, two trade bodies and a press operation to a single track.

What an indie can do

On Spotify, the route for a voice-impersonation claim is a legal report form with "Publicity / Likeness" selected. The claim has to come from the artist or someone acting for them, because Spotify says it can't tell a cleared sample or collaboration from an uncleared one. The page doesn't give a review time. Spotify's impersonation policy has allowed vocal impersonation only with the impersonated artist's authorisation since September 2025, and Spotify said then that it would cut wait times on its separate content-mismatch process.

The other tool is Artist Profile Protection. It is still in limited beta, and Spotify says it will reach all artists "as soon as we possibly can". MBW reports that the original version held releases until the artist approved them. In the current standard setting, an opted-in artist gets an email when music is delivered under their name and can remove themselves, but if they do nothing the release goes live on schedule. Holding releases for approval is now an advanced setting with limited access. Spotify's own page warns that it can delay or block legitimate releases.

Other independents have had a worse time than Moran. Danish musician Chris Minh Doky told DMN that the companies behind the fake releases weren't responsive, that bringing in a lawyer didn't settle it, and that what finally worked was shaming them publicly and asking fans to report the tracks. Six months later the fakes were back.

What we think

The 260,000 figure is being read as a sign of how much Sony is protecting its artists. We read it the other way. It shows that notice-and-takedown doesn't scale. The number of requests has grown faster than anyone can clear them, and the only parties who can keep up are the ones who can pay people to keep sending notices. Every one of those requests is work done after the fake has gone live and the release-week attention has already been taken.

That makes the default setting of Artist Profile Protection the decision that matters most to independents, more than any takedown total. A major's team can watch an inbox and remove fakes as they appear. A solo artist with a day job will miss the email, and under the current standard setting a missed email means the fake goes live. Holding a release until the artist approves it should be available to any artist who asks for it, not rationed as an advanced feature. The Artist Key is the useful idea here. Distributor EmuBands describes it as an identifier, generated in Spotify for Artists, that pre-approves a release automatically when it's included in the delivery. It moves the check to the point of delivery, which is where the fraud gets in.

The NO FAKES Act would give independents a federal right in their voice and likeness, which they don't clearly have now. MBW notes that cloning a voice isn't clearly copyright infringement in the US. The bill cleared the Senate Judiciary Committee unanimously on 18 June. It still works on the same model as Sony's operation, though: the rights holder flags a replica and the platform removes it. A right you have to police yourself is better than no right. But on its own it is still a system built for whoever can send notices at scale.

There is one thing Sony could do for everyone: publish how many of its 260,000 requests ended in a removal, how long removals took, and how many were clones rather than misdelivered tracks. Those numbers would show whether the system works, and independents can't produce them for themselves.

Sources

Sources

  1. Music that impersonates another artist's voice, Spotify for Artists Support (primary source). Accessed 23 Oct 2026.
  2. Introducing Artist Profile Protection: Giving Artists More Control Over What Shows Up Under Their Name, Spotify for Artists (primary source). Accessed 23 Oct 2026.
  3. Spotify Strengthens AI Protections for Artists, Songwriters, and Producers, Spotify Newsroom (primary source). Accessed 23 Oct 2026.
  4. Sony Music has now asked platforms to remove more than 260,000 AI deepfakes imitating artists like Harry Styles, Adele and Michael Jackson, Music Business Worldwide. Accessed 23 Oct 2026.
  5. Report: Sony Music doubles AI deepfake takedowns to more than 260,000, Music Week. Accessed 23 Oct 2026.
  6. Sony reportedly demanded music platforms take down over 260,000 AI deepfake tracks, The Verge. Accessed 23 Oct 2026.
  7. Spotify AI Hijacking Crisis Intensifies as Jazz Musicians Targeted, Digital Music News. Accessed 23 Oct 2026.
  8. NO FAKES Act passes Senate Judiciary Committee, Recording Academy (GRAMMY.com). Accessed 23 Oct 2026.
  9. Spotify Artist Profile Protection, EmuBands. Accessed 23 Oct 2026.

Researched with AI assistance, written and checked by WATMM Desk.

Subscribe to We are the music makers

One email a week: the music-tech and AI-in-music news that mattered, and what we think about it.

Sign-ups open soon.