The deal, announced on Thursday, comes as expanding AI capabilities blur the line between human and machine-made songs, enabling a new type of streaming fraud where bad actors upload thousands of AI-generated numbers designed to trigger algorithmic recommendations and siphon royalties away from artists and songwriters.
Deezer said it successfully identified and removed up to 85% of fraudulent AI-generated music streams from its royalty pool in 2025, flagging over 13.4 million AI tracks. The platform now receives around 60 000 fully AI-created tracks every day, roughly 39% of total daily uploads, up from 10% in January last year.
Deezer’s royalty pool comprises 70% of subscriber revenue, according to Chief Executive Officer, Alexis Lanternier. The detection tool analyses audio signals for patterns created by AI music generators such as Suno and Udio, identifying subtle anomalies inaudible to human ears. The company has trained the system on 94 million songs and filed two patents for the technology in 2024, it said.
However, Swedish royalty society Stim told Reuters that detection tools address only part of the problem and that technology cannot be an answer to issues surrounding underlying musical composition and copyright ownership. Stim believes mandatory licensing and full transparency for training data would prevent fraud at its source.
Deezer is in discussions with other European collective societies regarding licensing and plans to engage with organisations in Los Angeles during Grammy Week to expand its efforts, Lanternier said. Sacem did not reply to Reuters’ request for comment.
–Reuters–
