Sony Music Seeks Removal of Over 260,000 AI Deepfake Tracks

As of the end of September, Sony Music Entertainment had asked platforms to remove more than 260,000 AI-generated tracks impersonating its artists, according to figures supplied to the Financial Times and reported by Music Business Worldwide (MBW).

The tally is almost double the 135,000 tracks Sony had sought to remove by the end of March. Artists identified in the Financial Times report include Harry Styles, Adele, Britney Spears, Queen and Michael Jackson.

The figures count tracks Sony has asked platforms to take down, but do not establish how many were removed. MBW notes that the Financial Times did not identify the services receiving the requests or report how quickly they acted.

Sony said it expected platforms to act promptly and prevent the same content from reappearing. “But it’s becoming an uphill struggle,” the company told the Financial Times, according to MBW.

Sony’s tally concerns unauthorised artist impersonation. It does not measure all AI-generated music or establish that every flagged track was involved in streaming fraud.

In Australia, ARIA’s recent chart rule changes govern which recordings qualify for its charts. Under ARIA’s announcement, wholly AI-generated recordings are excluded, while AI-assisted recordings can qualify if they are substantially human-made and meet the other chart eligibility requirements. The rules took effect with the chart dated 31 August 2026.

ARIA’s accompanying FAQ makes the distinction more specific: a recording with an AI-generated lead vocal or key instrumental performance is ineligible. A human lead vocal with AI backing vocals can remain eligible, subject to the other chart requirements.

The FAQ also states that chart eligibility does not determine royalties, radio play or what a streaming service carries. Excluding a recording from the charts therefore does not itself remove it from a platform.

Spotify, meanwhile, says its impersonation policy permits vocal impersonation only when the artist has authorised it. Spotify separately identifies music fraudulently delivered under another artist’s name as an impersonation problem, whether the recording is AI-generated or human-made.

Spotify's optional Artist Profile Protection feature remains in limited beta. Its guidance, updated on 9 September, says artists can remove their names from wrongly attributed releases. Under the standard setting, a release goes live as scheduled if the artist takes no action.

A setting that blocks releases until an artist approves them has limited availability, Spotify says.

These policies do not establish whether Spotify received or complied with any of Sony’s requests. MBW’s account provides no Australian breakdown, so the scale of unauthorised artist impersonation locally remains unclear.