Deezer Adds Toggle to Filter Out AI-Generated Music

Deezer will introduce a user-controlled toggle before the end of 2026 that strips AI-generated tracks from its streaming interface, marking the first time a major music platform has handed listeners a dedicated switch to purge synthetic content entirely.

The feature arrives as algorithmic uploads overwhelm traditional catalog growth. More than half of all tracks currently uploaded to Deezer are fully generated by artificial intelligence, a sharp acceleration from 18 percent recorded just last year. The Paris-based company now processes roughly 90,000 machine-made songs daily, pushing its cumulative backlog of AI audio past 42 million tracks.

This volume reflects an arbitrage strategy exploiting streaming royalty structures. Because platforms like Deezer, Spotify, and Apple Music generally distribute revenue from a shared pool based on total stream counts, operators can upload thousands of low-effort, AI-generated tracks daily, banking on sheer mathematical probability that a fraction will accumulate enough passive plays—often through automated bot networks—to generate payouts. Deezer’s internal data indicates that up to 85 percent of streams attributed to AI-generated songs involve fraudulent activity, transforming synthetic uploads from a novelty into a direct financial drain on human artists whose per-stream royalties are diluted by inflated platform totals.

Consumer tolerance for the flood is eroding. Deezer’s proprietary research found that nearly half of its streaming users want AI music removed from their feeds completely, aligning with broader industry surveys showing listener sentiment sours proportionally as platforms become saturated with synthetic audio.

To power the new filter, Deezer is leveraging an AI detection system it developed internally and previously offered to license to rival services. While the company acknowledges the classification tool is not infallible, it maintains the false-positive rate—where a human-created track might be mistakenly flagged as synthetic—remains very low. Unlike standard metadata tagging, which relies on uploaders honestly self-reporting how a track was made, audio fingerprinting and spectral analysis attempt to identify the acoustic signatures typical of generative models such as Suno and Udio.

Competitors have so far opted for softer interventions rather than outright removal options. YouTube Music recently rolled out a labeling system that tags AI-generated tracks without altering their placement in recommendations or search results. Spotify applies similar disclosures to specific AI artist pages. Neither platform currently allows users to systematically exclude this content from algorithmic playlists or radio features.

Deezer holds a distinct structural advantage in attempting aggressive filtering. As a mid-tier global service with approximately 10 million subscribers—dwarfed by Spotify’s hundreds of millions—it faces less immediate pressure to maximize raw catalog size for investor metrics and can position itself as a curator-first alternative. The French government has also moved faster than U.S. regulators on AI transparency; a 2024 French law requires clear labeling of AI-generated cultural content, providing Deezer with a domestic regulatory framework that already anticipates these exact disclosure mechanisms.

The toggle will roll out across Deezer’s mobile and desktop applications globally before December 31.