Spotify Says It Stripped 50 Million AI Slop Tracks From Recommendations
The platform is tightening its recommendation algorithm and adding an "AI Persona" badge, even as it stops short of labeling every AI-made song.
Spotify says it has scrubbed more than 50 million AI-generated slop and spam tracks from user recommendations over the past year, according to an announcement the company made Tuesday. The streamer also confirmed it has retooled its recommendation system to push more music from human artists to the surface.
The move addresses a problem that's been building for a while: AI tools have made it trivially easy to flood streaming platforms with mass-produced, low-effort tracks designed to game algorithms and rack up royalty payouts rather than to be, you know, listened to. Spotify's editorial team isn't exempt from the crackdown either. The company says its editors "hold the same standard and don't program content identified as slop or spam," and that listener complaints about this kind of content turning up in recommendations have dropped noticeably since the changes went into effect.
To catch the junk, Spotify says it leans on a mix of signals: how listeners engage with and react to tracks, patterns in how songs get uploaded, metadata, and third-party AI detection tools. It's a cruder approach than an outright AI label, but it's meant to work behind the scenes to keep spam out of the discovery pipeline without necessarily tagging every track that used AI somewhere in its production.
That distinction matters, because Spotify isn't requiring labels on all AI-made music. Instead, it's rolled out something narrower: an "AI Persona" badge for artist profiles built around an AI identity rather than a real human. Spotify won't recommend music tied to those AI personas, drawing a line specifically around fabricated artist identities rather than AI use in general.
A tougher landscape across streaming
Spotify's stance looks relatively light-touch next to what other platforms are doing. YouTube Music has started slapping labels directly on AI-generated songs. Deezer gives listeners a toggle to filter AI slop out of their experience entirely. Bandcamp has gone furthest, banning AI music from its platform outright. Each of those represents a different bet on how much transparency, or restriction, listeners actually want. Spotify, for now, is betting that cleaning up recommendations quietly works better than forcing disclosure everywhere.
The scale of the spam problem Spotify is describing isn't new, but it is growing. Last year, the company said it had taken down more than 75 million songs identified as spam. This year's 50 million figure for AI slop and spam pulled specifically from recommendations suggests the issue hasn't gone away, even as Spotify builds out detection tools to deal with it.
At the same time, Spotify is leaning into AI rather than rejecting it wholesale. The company is partnering with Universal Music Group and independent music publisher Believe to launch a tool for creating AI-powered music remixes, a sign that Spotify sees a difference between AI as a creative tool used by legitimate artists and labels, versus AI as a mechanism for churning out anonymous, algorithm-gaming filler.
That split, policing slop while building AI remix tools with major labels, captures where the whole industry seems to be landing. Nobody's trying to ban AI from music outright except Bandcamp. The real fight is over which uses of the technology get rewarded and which get filtered out before a listener ever hits play.
