Article · AI translation
IT Home reported on October 21 that Spotify announced on Tuesday local time that, over the past year, it had removed more than 50 million songs identified as low-quality AI-generated spam or spam-streaming tracks from users’ recommendation feeds. It also adjusted its recommendation system to prioritize music by human creators.
Spotify said: “The platform’s editorial team applies the same standards and will not include content deemed low-quality AI spam or spam tracks in recommendation feeds.” “After these changes, listener complaints about such content appearing in recommendation feeds declined markedly.”
According to IT Home, Spotify does not require all AI tracks to be labeled, but it has introduced an “AI Persona” label for creators who use virtual AI identities rather than being real people, and it will not recommend music created by such AI personas. Other music streaming platforms have adopted stricter policies on AI music: YouTube Music labels AI songs; Deezer lets users filter out low-quality AI music themselves; and Bandcamp has banned AI music from being listed altogether.
Spotify said it uses a combination of users’ listening behavior and feedback, upload characteristics, metadata, and third-party AI detection tools to determine whether a track is low-quality AI-generated spam. The company said last year that it had removed more than 75 million songs identified as spam. Meanwhile, Spotify is working with Universal Music Group and independent music distributor Believe to launch a tool for creating AI music remixes.