Deezer says 44% of newly uploaded music on its platform is AI-generated, equal to roughly 75,000 tracks per day. The company also said most of the listening activity linked to those tracks comes from fraudulent bot-driven streams, raising pressure on a royalty system that was built around human creators and real audiences.
AI tracks and bot traffic are growing at the same time
The figures point to a larger problem for streaming platforms: a rising share of new catalog growth is no longer tied to traditional music production. Deezer said some AI-made tracks are uploaded at scale and then paired with automated plays, creating inflated activity made up of machine-generated content and machine-generated listening. In practical terms, that means royalty pools can be diluted by streams that do not reflect genuine fan demand.
Deezer says its in-house detector keeps false positives below 0.01%
To address the issue, Deezer developed its own AI content detection system and said the false-positive rate stays below 0.01%. By the company’s measure, that means fewer than 1 human-made track would be incorrectly flagged out of every 10,000 tracks labeled as AI. Deezer is trying to push for clearer disclosure standards around AI music, though labeling alone does not solve the deeper problem of how platforms identify and reward legitimate activity.
Listener survey found 97% could not tell AI songs apart
In a survey commissioned by Deezer, participants listened to three songs in a row, two of which were AI-generated. The result was stark: 97% of listeners could not correctly identify which song came from AI and which came from a human artist. That gap helps explain why fake streaming can spread so easily. If listeners cannot reliably hear the difference, manipulated play counts are harder to separate from normal listening behavior.
Pressure builds on larger platforms that do not label AI music
Deezer’s public stance also puts attention on the approach taken by bigger streaming services. According to the source material, Spotify and YouTube Music do not currently apply proactive labels to AI-generated music. A report cited in the article said that last June as many as 500,000 Spotify users were repeatedly listening to AI-generated “bands” without knowing it. Those artist pages looked like standard musician profiles, complete with cover images and bios, but carried no AI marker.
The dispute is now centered on authenticity in streaming payouts
At the core of the dispute is a simple economic imbalance. AI music can be produced at near-zero cost and uploaded with little friction, while royalty systems have not been redesigned to block abuse tied to synthetic content. The source argues that if users can generate hundreds of tracks in a day and use bots to manufacture streams for royalty payouts, the line between real listening and fabricated activity starts to erode. Deezer’s disclosure brings that conflict into public view.