THE CRUNCH

Spotify says it removed more than 50 million tracks it categorised as AI-generated slop or spam from user recommendations over the past year, and has tweaked its recommendation system to promote more music from human artists. The company says editors no longer include tracks identified as slop or spam in curated playlists, and that it has seen a marked decrease in listener complaints since making the changes. Detection relies on a mix of listener feedback, upload patterns, metadata and third-party AI detection tools.

There is an important catch: the tracks have been pulled from recommendations, not necessarily wiped from the platform entirely. Spotify has also not updated its separate figure of more than 75 million spammy tracks removed for rule violations, the same number it gave in September 2025 when it announced its AI policy changes, though a spokesperson told Music Business Worldwide that deletions have continued.

The company is adding disclosure tools alongside the crackdown. Artists are submitting tens of thousands of AI credits per day, visible in the mobile Now Playing view, and thousands of virtual artists now carry an AI Persona label. Tracks from these AI personas are excluded from editor-curated playlists and, by default, from recommendations. Spotify also says its 1,000-stream minimum before royalties are paid helps, with tracks flagged as likely AI-generated ten times more likely to fall below that threshold.

WHAT HAPPENS NEXT

Spotify says it will need to keep updating its systems as spammers change tactics to evade the rules, and Deezer's full user filter for AI-generated audio is due by the end of the year.