Spotify says it removed 75 million AI-generated spam tracks

Spotify said it had removed 75 million tracks from its catalogue that it classified as spam, and announced new policies banning impersonation through AI voice cloning, prohibiting mass-uploaded spam submissions, and targeting tracks created to generate fraudulent royalty payments.

The company stopped short of banning AI-generated music itself. Instead of automatically detecting and excluding fully AI-generated tracks — the approach taken by rival Deezer, which said it was filtering such tracks out of its recommendation algorithms and withholding royalties from them, and estimated that around 28% of the roughly 150,000 tracks submitted to streaming services daily were purely AI-generated — Spotify said it would work with the music industry body DDEX to build a shared standard for disclosing which elements of a track were AI-generated, and planned to surface that information to listeners. It did not commit to using the disclosure data to affect royalty payouts or algorithmic promotion.

The announcement placed Spotify’s response to AI-generated content well behind Deezer’s more aggressive filtering, reflecting the streaming industry’s uneven and still largely voluntary approach to a problem — catalogues being flooded with low-cost synthetic music, some of it designed purely to extract royalties — that platforms had been slow to address publicly. The scale of the removal, at 75 million tracks, gave one of the first concrete public figures for how much of a major platform’s catalogue had been affected.