Spotify says it has excluded over 50 million tracks from recommendations this year as part of measures addressing artificial intelligence (AI) abuse, streaming fraud and music discovery.
The company disclosed the figure in a one-year progress update. Separately, it says it removed over 75 million tracks during the past year.
Restrictions on recommendations
The 50 million tracks identified as “slop and spam” have not all been deleted from Spotify’s catalogue. They have been excluded from personalised recommendations, including Discover Weekly and Release Radar, and from editorial playlists.
Spotify uses signals including listener reports and skip rates to identify tracks for these restrictions. Its policy permits AI-created music while targeting unauthorised voice clones, mass-generated spam and automated streaming.
Legitimate streams of synthetic tracks remain eligible for standard royalties. The measures aim to prevent royalty fraud and limit manipulation of personalised recommendations.
Labels for synthetic artists
Spotify has introduced two labels for artist profiles that do not represent a real person:
- AI Persona
- Likely AI Persona
Profiles carrying either label are automatically excluded from algorithmic recommendation channels. Spotify uses uploader self-disclosure to label AI artist profiles and excludes fully synthetic tracks from editorial playlists.
Under its Verified by Spotify standards, human artists can retain verification while using AI in their creative process, provided they represent themselves authentically.
Entirely synthetic personas are ineligible for the blue verification checkmark.
“As AI makes music more abundant, we believe listeners will value human artistry even more. Our job is to make Spotify the place they find it.”
Spotify
Credits disclose AI contributions
Listeners can access AI Credits through song details in Spotify’s mobile app. Participating music distributors submit the information, allowing creators to identify specific AI contributions.
Disclosures can cover AI-assisted lyric writing, voice models or instrumentals, rather than applying a single AI designation to an entire song.
