Sony Music Tops 260,000 AI Deepfake Takedown Requests

Sony Music had asked digital platforms to remove more than 260,000 AI-generated tracks impersonating its artists by the end of September 2026, nearly double the 135,000 it had flagged by the end of March, the Financial Times reported on Monday, 5 October. The deepfakes imitate artists including Harry Styles, Adele, Britney Spears, Queen and Michael Jackson, and Sony told the paper that keeping up is becoming an uphill struggle.

The new tally, which Music Business Worldwide and Music Week both picked up the same day, is the clearest measure yet of how fast voice-cloned songs are spreading across streaming services.

How Sony’s deepfake takedown count doubled

The growth is steep. Music Business Worldwide calculated that the new figure implies Sony sent more than 20,000 deepfake takedown requests a month on average between April and September. Sony described the problem as generative AI deepfake tracks that imitate its artists’ voices and likenesses without permission, and said it is happening at scale, harms artists, misleads fans and undermines the ethical use of AI.

In a statement quoted by Music Week, Sony said it would keep enforcing its artists’ rights and expects platforms to act quickly and to stop the same content from reappearing, but added that the work is becoming an uphill struggle. Music Business Worldwide noted that the report did not say how many of the flagged tracks platforms have actually removed, how quickly they acted or which services received the requests.

The longer history shows the curve. According to Music Business Worldwide, Sony executive Dennis Kooker told a US Senate AI forum in November 2023 that the company had sent close to 10,000 takedowns. By March 2025, Sony said it had requested more than 75,000 in a submission to a UK government consultation on AI and copyright, and by March 2026 the count had passed 135,000. Speaking at an IFPI event in London in May, Kooker said the problem with deepfakes is that they are demand-driven, according to Music Week, appearing when an artist is promoting new music and trading on the attention that campaign creates.

The streaming fraud problem behind the numbers

Deepfakes are only one part of a wider fraud problem. Music Business Worldwide, summarizing the original report, said fraudsters also use AI to mass-produce tracks and then deploy bots to stream them, diverting royalties that would otherwise go to artists and rights holders. Label executives estimate streaming fraud could cost the industry up to $2.2 billion a year, according to that summary.

Kooker has said nobody knows the true scale. He told Music Week in May that somewhere between 5% and 10% of streams could be fraudulent, and that the way the ecosystem works gives bad actors a strong incentive to manufacture streams. Music Business Worldwide added that a registry report cited in the coverage found 92.8% of 1,743 verified AI track submissions had been made with Suno, with one distributor handling more than three-quarters of those where distributor data was available.

The tension with AI music companies is not new. PromptCrates has covered how Sony and Warner sued Anthropic over Claude training data, how Suno shipped v6 music models with Warner, BMG and Believe, and how EDM producers began hunting AI music grifters on Suno.

What platforms and lawmakers are doing about deepfakes

The industry is trying to respond on several fronts. Music Week reported that Sony, Universal and Warner were among the companies backing the IFPI’s Streaming Integrity Initiative, an industry commitment against streaming fraud launched last month. Digital Music News noted that Universal Music is separately suing the distributor DistroKid, which Digital Music News said is not taking part in the initiative, and that Warner is currently the only major label with a Suno partnership.

Streaming services have tightened their rules too. Music Business Worldwide recounted that Spotify introduced an impersonation policy in September 2025 allowing vocal impersonation only when the impersonated artist has authorized it, and began piloting an opt-in Artist Profile Protection feature in March 2026. Spotify has since revised that beta so that, in its standard setting, artists are notified when music is delivered under their name and can remove themselves from incorrect releases.

Lawmakers are moving more slowly. The proposed US NO FAKES Act, which Sony supports, would create a federal right in a person’s voice and visual likeness. To keep their safe harbor, platforms would have to remove a flagged replica as soon as is technologically and practically feasible. Music Business Worldwide reported that the bill cleared the Senate Judiciary Committee unanimously on 18 June, but it has not become law, and cloning an artist’s voice does not clearly amount to copyright infringement in the US today.

What artists and AI music users should expect

For artists, the takeaway is that label enforcement is scaling but still reactive. Sony is filing requests faster than ever, yet its own description of an uphill struggle suggests takedowns alone cannot keep pace with how cheaply AI tools can generate soundalike tracks. Expect labels to push harder for upload-time detection, re-upload blocking and distributor accountability rather than relying on notices after songs go live.

For creators using AI music tools legitimately, the pressure will show up as stricter checks. Distributors and streaming services are likely to ask more questions about vocals that resemble well-known artists, and releases that imitate a recognizable voice without permission are increasingly likely to be pulled. Anyone experimenting with AI vocals should build on original voices or licensed models and keep records of how a track was made.

What to watch next is whether Sony publishes how many of the 260,000 tracks were actually removed, whether the Streaming Integrity Initiative produces measurable results, and whether the NO FAKES Act reaches a floor vote. If the monthly pace keeps rising, the next tally will show whether enforcement is gaining any ground.

• Financial Times: Sony Music steps up fight against streaming fraud
• Music Business Worldwide: Sony Music has now asked platforms to remove more than 260,000 AI deepfakes
• Music Week: Sony Music doubles AI deepfake takedowns to more than 260,000
• Digital Music News: Sony Music AI deepfake takedowns top 260K
• PromptCrates: Sony and Warner sue Anthropic over Claude training data
• PromptCrates: Suno ships v6 music models with Warner, BMG and Believe
• PromptCrates: EDM producers hunt AI music grifters on Suno