Spotify has taken more than 50 million tracks it identifies as spam or AI-generated 'slop' out of recommendations in 2026, the company announced on 20 October. The move limits which recordings it promotes to listeners; removing tracks from the service is a separate enforcement action.
The company also cites more than 75 million spam tracks removed from its catalogue, a figure first disclosed in September 2025 for the preceding 12 months. A spokesperson tells Music Business Worldwide that removals have continued since then, but Spotify is not releasing an updated total.
Spotify Combines Listening Signals With Upload Patterns
The recommendation changes follow an examination of tracks listeners skip, ignore or say they do not want to hear. Spotify says those recordings consistently overlap with content it identifies as spam or slop, prompting updates over the past six months. Its editors also avoid programming content given those classifications.
Listener behaviour is only part of the process. The system combines engagement and feedback with upload patterns, metadata and information from several third-party AI detection tools.
'No single signal catches slop or spam on its own', the company says in its update. It also looks for mass uploads, duplicate recordings, manipulation of search visibility, spammy playlist networks and artificially short tracks.
These checks extend beyond trying to identify how a recording sounds. Spotify acknowledges that AI detection has limitations and will become harder as generative technology develops. It reports a decrease in listener concerns since the recommendation changes, but supplies neither a measure of that improvement nor detection error rates.
Identity Badges Do Not Describe How Music Is Made
Listeners also encounter AI Persona badges, which identify artists presenting themselves as photorealistic AI-generated people. These labels concern the public identity attached to the music, rather than establishing how the recording was produced.
'The AI Persona badge is about the identity you present as an artist, not whether you use AI to make your music.' That distinction, set out in the artist guidance, allows a badged identity to release human-made recordings, AI-generated music or a mixture.
Listeners can still choose to play and follow those artists. Their music is excluded from editorial and personalised recommendations by default, but intentional engagement, such as following an artist, can create an exception.
AI credits provide a separate view of the recording itself. Artists submit them through participating distributors to identify generated contributions such as lyrics, vocals or production, rather than labelling an entire track as AI-made. The credits are in beta and visible on mobile; because disclosure is optional and distributor support varies, their absence does not establish that no AI was used.
Artists Can Challenge a Spotify-Applied Identity Badge
An artist receives a notification in Spotify for Artists when the service applies a 'Likely AI Persona' badge. That label reflects the platform's determination, rather than the artist's own disclosure.
The written guidance gives artists an explicit option: 'Appeal the decision if you believe the badge was applied in error.' Instructions are available inside Spotify for Artists, and the badge remains if the artist takes no action.
Artists cannot remove a Spotify-applied badge themselves. The company reviews the available information and makes the final decision; the public guidance does not promise a response time or a particular outcome.
That process addresses a classification with consequences for how an artist's music reaches listeners, but its scope is specific. The published badge instructions do not establish a general procedure for challenging ordinary spam-filter exclusions.
An artist seeking review of a track's recommendation status therefore cannot treat the identity-badge appeal as a documented route for that separate decision.
