SAMPLEDWhy Has Sony Music Asked to Remove 260,000 AI Deepfake Songs?
Sony's takedown requests for AI tracks faking its artists nearly doubled in six months. Here's what's driving it and how independent artists can protect their name.
By the end of September 2026, Sony Music had asked streaming and digital platforms to take down more than 260,000 AI-generated tracks pretending to be its artists, the Financial Times reported on October 5. That's almost double the 135,000 it had flagged by the end of March.
Sony's own description of the situation: "It's becoming an uphill struggle."
What exactly is Sony taking down?
Sony describes them as "generative AI deepfake tracks that imitate our artists' voices and likenesses without permission." According to the FT, the artists being faked include Harry Styles, Adele, Britney Spears, Queen and Michael Jackson.
Sony's numbers don't separate two different problems that platforms like Spotify treat differently:
- AI voice clones: songs made to sound like a real singer who never recorded them.
- Profile hijacking: music (AI or not) delivered to another artist's profile so it shows up next to their real releases.
How fast is the problem growing?
Sony's takedown count has climbed every time it has shared a figure:
| When | AI deepfake takedown requests |
|---|---|
| November 2023 | close to 10,000 |
| March 2025 | more than 75,000 |
| March 2026 | more than 135,000 |
| End of September 2026 | more than 260,000 |
Between April and September, that works out to an average of more than 20,000 requests a month, according to Music Business Worldwide (opens in a new tab). The FT didn't say how many of the tracks were actually removed, how fast platforms acted or which services got the requests.
Why are deepfakes spreading so quickly?
Sony executive Dennis Kooker told the BBC in March that deepfakes are "a demand-driven event" that take advantage of an artist being out there promoting their music. When fans are searching for a new single, fakes ride that attention.
Money is the other driver. The FT reports that fraudsters also use AI to mass-produce tracks and then use bots to stream them, taking royalties that would otherwise go to real artists. It's the same problem behind Universal's lawsuit against DistroKid and Qobuz's new AI labels.
Photo: Claus Grünstäudl via Wikimedia Commons (opens in a new tab) (CC0)
What are streaming services doing about it?
Spotify introduced an impersonation policy in September 2025 that says vocal impersonation is only allowed "when the impersonated artist has authorized the usage." In March 2026 it began testing Artist Profile Protection, an opt-in feature. In its standard setting, artists are notified when music is delivered under their name and can remove themselves from releases that aren't theirs. Holding releases until the artist approves them is now an advanced setting with limited access.
In the US, the proposed NO FAKES Act would cover "digital replicas," which it describes as computer-generated copies that are readily identifiable as a person's voice or likeness. It hasn't become law.
Can this happen to independent artists?
Yes. Sony can afford to file 20,000 takedowns a month; most independent artists can't. But you don't need a label's legal team to protect yourself:
- Turn on profile alerts. If you're on Spotify for Artists, opt in to Artist Profile Protection so you hear about any release delivered under your name.
- Search yourself regularly. Look up your artist name on Spotify, Apple Music, YouTube and TikTok every few weeks, especially around a release.
- Report fakes quickly. Use each platform's content mismatch or impersonation report, and tell your distributor right away.
- Keep proof of your work. Save your session files, stems and dated demos so you can show what's really yours.
- Know your rights. Our explainer on whether AI music is legal covers where the law stands now.
The bottom line
Sony's figures show AI deepfakes are growing faster than the major labels can take them down. For independent artists, the practical defense is the same as for the labels, on a smaller scale: watch your profiles, report fakes fast and keep proof of your real work.