Spotify removes 50 million AI slop and spam tracks from recommendations

Spotify removes 50 million AI slop and spam tracks from recommendations

Spotify says it has removed more than 50 million songs considered AI-generated slop or spam from user recommendations over the past year. The figure comes from an announcement made on Tuesday, which also says Spotify has changed its recommendation system to highlight more music made by human artists. The wording matters: the tracks were taken out of recommendations, and the source does not say they were deleted from the platform.

Spotify says it decides whether a track may be AI-generated slop using a combination of listener engagement and feedback, upload patterns, metadata, and third-party AI detection tools. Its editors follow the same line. In Spotify's words: "Our editors hold the same standard and don’t program content identified as slop or spam." The company adds that since the changes it has seen "a marked decrease in listener concerns about this content showing up in recommendations." No figure is given for that decrease.

Spotify has stopped short of mandating labels on all AI tracks. Instead it launched an "AI Persona" badge for artists that use an AI identity and do not represent a real person, as The Verge describes it. Spotify also will not recommend music from AI personas.

The Verge sets this against rivals that are taking a tougher stance on AI music: YouTube Music puts labels on AI songs, Deezer gives users an option to filter out AI slop, and Bandcamp bans AI music entirely.

There is earlier context. Last year the company said it took down more than 75 million songs identified as spam. That is a separate, earlier figure, and the source does not say whether it overlaps with the new 50 million. Spotify is also partnering with Universal Music Group and the independent music publisher Believe to launch a tool for creating AI music remixes; the source gives no detail on how that tool works or when it launches.

Key facts

  • Spotify says it removed more than 50 million songs considered AI-generated slop or spam from user recommendations over the past year.
  • Spotify changed its recommendation system to highlight more music by human artists, and its editors do not program content identified as slop or spam.
  • A new "AI Persona" badge marks artists that use an AI identity and do not represent a real person; Spotify will not recommend music from AI personas.
  • Spotify detects slop using listener engagement and feedback, upload patterns, metadata, and third-party AI detection tools.
  • Spotify has not mandated labels on all AI tracks, while YouTube Music labels AI songs, Deezer offers a filter, and Bandcamp bans AI music.

Why it matters

AI-generated filler is a growing complaint on streaming platforms, and Spotify is one of the largest. Its answer is to keep such tracks out of recommendations and to push human artists up, rather than to ban AI music or label every AI track. Spotify says listener concerns about this content showing up in recommendations have dropped markedly since the changes. It also separates AI personas from real artists with a badge. The 50 million figure follows an earlier claim that the company took down more than 75 million spam songs last year.

Who it affects

Listeners are the first group: Spotify says it has seen a marked decrease in listener concerns about slop and spam showing up in recommendations. Human artists stand to gain, since the recommendation system now highlights more of their music. Artists that use an AI identity and do not represent a real person get the "AI Persona" badge and will not be recommended. Uploaders of spam or slop lose recommendation reach. Rival platforms are the comparison point: YouTube Music, Deezer and Bandcamp each handle AI music differently.

How to use it

There is nothing for a listener to switch on in the source; the changes sit in Spotify's recommendations and editorial programming. What a listener can look for is the "AI Persona" badge on artists that use an AI identity without representing a real person. Listeners who want a stricter filter are pointed by the source to other services: Deezer gives users an option to filter out AI slop. The source gives no pricing or tier details.

How solid is it

The numbers are Spotify's own, reported by The Verge from the company's Tuesday announcement; no independent check is cited. The 50 million figure covers songs removed from recommendations over the past year, not tracks deleted from the platform. The claim of a marked decrease in listener concerns is unquantified. The source gives no calendar date for the announcement, only "Tuesday".

Risks and caveats

Spotify says it judges a track using engagement and feedback, upload patterns, metadata and third-party AI detection tools. It describes these as a way to determine whether a track "may be" AI-generated slop, so the call is probabilistic. Without mandatory labels on all AI tracks, AI music that is not flagged as slop or spam is not marked, and the badge only covers AI personas. The source does not say whether the 50 million overlaps with the earlier 75 million, nor give the share of AI tracks in the catalog. Spotify's partnership with Universal Music Group and Believe on an AI music remix tool shows the company is not rejecting AI music outright; the source says nothing more about that tool.

“Our editors hold the same standard and don’t program content identified as slop or spam”

— Spotify