[DigitalToday reporter Jae-won Choi] Global music streaming platform Spotify has excluded more than 50 million AI-generated low-quality tracks and spam audio from its recommendation lists. It revamped its recommendation system so that works by real musicians are shown more often in response to the spread of AI content.

On Oct. 20, local time, IT outlet The Verge reported that Spotify announced it had excluded more than 50 million low-quality and spam tracks from its recommendation system since the start of this year.

Spotify improved its recommendation algorithm by analysing users' listening history and feedback over the past six months. It said it found characteristics of low-quality content in tracks that users repeatedly skipped or refused to listen to.

Spotify is detecting mass uploads, duplicate tracks, manipulation of search exposure and spam playlists. It identifies low-quality content by combining listener responses, upload patterns, track information and external AI detection tools.

The same standards also apply to editor-curated playlists. Spotify said user complaints about low-quality music being surfaced have clearly declined since the overhaul of its recommendation system.

It is not blocking all music made with AI. Spotify is attaching an AI persona badge to accounts that use a virtual artist identity.

Music by AI artists that are not real people is excluded from recommendations by default. Exceptions may apply, including when users directly follow the artist.

Spotify also operates an AI credits feature that discloses whether AI was used. If artists reveal what parts of a song's production involved AI, users can check the related information.

Separately, Spotify said it deleted more than 75 million spam tracks from the platform for serious or repeated rule violations.

Platform responses to AI music vary. YouTube Music added labels to AI-generated tracks, and Deezer provides an AI track filtering feature. Bandcamp banned AI music.

Spotify is also developing AI-based remix and cover creation tools in cooperation with music companies including Universal Music Group and Believe.

It is seeking a balance between using the technology and protecting artists by restricting low-quality AI content while supporting AI music production with creators' consent.