Deezer plans to let listeners remove fully AI-generated music from their search experience before the end of 2026. The global toggle will hide synthetic tracks, albums and related artists, while allowing users to reverse the setting whenever they want.
From detection to consumer control
The feature builds on a detection system Deezer introduced in 2025. The company said it identified and excluded more than 42 million AI-generated tracks from algorithmic and editorial recommendations between January 2025 and August 2026. It has also tagged more than 14 million albums.
The scale of incoming content explains the product decision. Deezer said it receives nearly 90,000 fully AI-generated tracks per day, representing more than half of daily deliveries on peak days. When supply grows that quickly, recommendation quality, fraud controls and royalty allocation become operating issues rather than abstract debates about creativity.
Demand for a filter is also measurable, although the evidence comes from company-sponsored research. A November 2025 Deezer and Ipsos survey found that 45% of streaming users wanted the ability to filter out fully AI-generated music. The same research said 80% believed such music should be clearly labeled. Those results support user demand, but they do not prove that the new toggle will increase subscribers or reduce churn.
A strategic feature for a smaller platform
Deezer competes against much larger global services, so a transparent AI policy can help it differentiate without matching every rival on scale. The feature gives listeners a direct choice and gives artists a clearer signal about how synthetic content enters recommendations.
The financial test is whether that positioning improves direct-subscriber growth or retention. Deezer reported first-half 2026 revenue of €268 million. Direct subscribers in France reached 3.9 million, up 8.4%, while subscribers in the rest of the world grew 9.2%. Management maintained its goal of roughly flat full-year revenue versus 2025 while delivering positive adjusted EBITDA and free cash flow.
The trade-off is detection accuracy. False positives could hide legitimate work, while false negatives would weaken trust in the filter. Deezer says its system can generalize beyond specific models, but independent error rates were not available in the materials reviewed.
The rollout creates a concrete measurement period. Investors should watch adoption of the toggle, changes in search and recommendation engagement, fraud-related costs, subscriber retention and any artist disputes. The feature is strategically coherent; its financial value will depend on whether users care enough to change behavior or choose Deezer over alternatives.