Myth check: "Spotify bans AI music"

The claim: Spotify bans AI music, so anything made with AI will be taken down or never get on the service.

Verdict: false. Spotify's own policy post says it treats music equally whatever tools made it. What it restricts is narrower: spam tactics, unauthorised voice clones, and how AI-only artist profiles are recommended. The claim mixes those up into a ban that does not exist.

Where it spread

The ban framing grows out of headlines about removals and limits rather than a ban. In September 2025 Spotify said it had removed more than 75 million "spammy" tracks, and coverage often turned that into "AI tracks removed". In August 2026 blue News reported that AI Persona profiles would be "systematically removed from all automatic playlist recommendations and radio streams", saying algorithmic visibility "will effectively be set to zero for AI musicians". Read quickly, "removed" becomes "banned". And when a big platform did ban AI music, it was Bandcamp: SoundGuys ran "Bandcamp bans AI music outright, while Spotify struggles" on 15 Jan 2026, which some readers then pinned on the wrong service.

What Spotify's policy actually says

Spotify's 25 Sep 2025 post has three parts, and none is a ban.

• Impersonation. "Vocal impersonation is only allowed in music on Spotify when the impersonated artist has authorized the usage." That is a ban on unauthorised voice clones, not on AI.
• Spam filter. A filter that tags uploaders and tracks using mass uploads, duplicates and artificially short tracks, and stops recommending them. The target is the tactic, whoever or whatever made the audio.
• AI disclosures. Spotify backs a DDEX standard for credits that say where AI was used (vocals, instrumentation, post-production) and shows them in the app. The post says this is "not about punishing artists who use AI responsibly" and not about down-ranking tracks that disclose.

The later AI Persona change, as blue News reported it, keeps those tracks on the platform and findable in search; it cuts recommendations, and it says a human artist "does not automatically become an AI persona just because AI is used as a creative tool". We covered the badge itself in our AI Persona piece.

What we think

"Banned" is the wrong word, and the wrong word leads to bad decisions. An artist who believes AI is banned may hide AI use, which is exactly what the disclosure credits are meant to avoid, and hiding it is a worse bet than disclosing it (see how to disclose AI use through DDEX). An uploader who believes AI is fine everywhere misses the real line: mass uploads, short tracks and fake voices will get filtered whether or not AI was involved.

The accurate summary is that Spotify allows AI music but is turning down the volume on it: the spam filter and Persona rules cut how much of it gets recommended, which for most listeners is how music gets found. In practice, being off recommendations can hurt nearly as much as a takedown, so calling it "a ban" is wrong in law but makes some sense from the uploader's side. Our reading is that Spotify chose this middle path on purpose: it keeps the catalogue and the licensing simple, and puts the pressure on discovery instead. For the scale of what it has pulled from recommendations this year, see Spotify's 50 million slop and spam tracks, and for whether the catalogue is "mostly AI", our earlier myth check.