Deezer will roll out a setting worldwide before the end of 2026 that lets listeners keep tracks it classifies as fully generated by artificial intelligence from appearing in search results. The company already excludes detected fully synthetic tracks from editorial playlists and algorithmic recommendations.
The distinction matters: Deezer says the setting applies to fully AI-generated music, not music that uses limited AI assistance within a broader human creative process. Users will be able to switch the filter off and restore those tracks, albums and related artists to search.
Deezer says its detection system identified and excluded more than 42 million AI-generated tracks from recommendations between January 2025 and August 2026 and has tagged more than 14 million albums. The company reports receiving nearly 90,000 fully AI-generated tracks a day, exceeding half of all deliveries on peak days, although those tracks account for only 1% to 3% of listening.
Those figures and the performance of Deezer’s patent-pending detection system are company claims that have not been independently audited. Deezer says as much as 85% of 2025 listening attributed to fully AI-generated tracks was fraudulent and that manipulated streams were excluded from royalty payments.
A Deezer-commissioned Ipsos survey published in November 2025 questioned 9,000 people in the United States, Canada, Brazil, the United Kingdom, France, the Netherlands, Germany and Japan. Deezer says 45% of streaming users wanted the ability to filter out fully AI-generated music.
The move makes automated classification a direct gatekeeper for music discovery. Deezer has not announced the precise launch date, a public error rate or what notice and appeals process will be available if a human creator believes a track, album or artist has been wrongly classified.
Sources: Deezer’s October 13 announcement and Music Business Worldwide’s independent industry report.
Radio News Now question: What evidence and appeal process will Deezer provide to artists and distributors when its system classifies their work as fully AI-generated?