The French streaming platform says 44% of daily uploads are AI-created, a figure that crystallizes the existential pressure bearing down on working musicians and the royalty structures that sustain them.
Deezer has put a number to something the music industry has been anxiously circling for years. Nearly half of every song uploaded to its platform on any given day is now AI-generated, the company disclosed, marking what may be the clearest signal yet that artificial intelligence has moved from a novelty to a structural force reshaping how recorded music is produced, distributed, and monetized.
The 44% figure is striking not just for its size but for what it represents in practice. Deezer CEO Jeronimo Folha and other executives have spoken previously about the platform's struggle with so-called functional or noise content , tracks assembled without human artistic intent and uploaded in bulk specifically to harvest streaming royalties. That category, once a nuisance, now accounts for nearly half of new content entering the pipeline each day.
Streaming economics make this more than an aesthetic concern. Platforms like Deezer distribute royalties from a shared pool, which means every AI-generated stream that earns a fraction of a cent does so at the direct expense of a human artist somewhere else in the system. If AI-generated tracks capture even a modest slice of total listening time, the per-stream rate for working musicians compresses further , and that rate is already a persistent grievance across the industry.
Deezer has previously proposed restructuring its royalty model to deprioritize noise content and more directly reward professional artists. That proposal, which drew attention when it was floated, now has fresh urgency attached to it. The 44% disclosure is likely to accelerate those internal deliberations and force the conversation onto competitors who have so far avoided publishing comparable data about their own upload volumes.
Industry Pressure Is Building
Universal Music Group, Sony Music, and Warner Music Group have all taken public positions demanding that streaming platforms adopt clearer policies on AI content, and industry lobbying bodies have pushed for regulatory frameworks requiring disclosure or labeling of AI-generated tracks. Until now, those calls have lacked the kind of concrete, platform-level data that makes legislative action feel urgent. Deezer's announcement changes that calculus.
The tools driving this upload surge are not obscure. Platforms like Suno and Udio have made it possible for anyone to generate a convincing, studio-quality track in seconds, and the economics of streaming create a rational incentive to upload at scale regardless of whether anyone ever meaningfully listens. It is a pure arbitrage play on the royalty pool, and it has clearly reached an industrial scale.
What remains an open question is whether Spotify and Apple Music face similar proportions inside their own pipelines. Both platforms are significantly larger than Deezer, and if anything close to 44% of their daily uploads follow the same pattern, the aggregate dilution effect on artist earnings would be substantially more severe. Neither company has volunteered equivalent figures, and the pressure on them to do so will intensify now that Deezer has made the first move.
The recorded music industry has survived technological disruption before , digital downloads, peer-to-peer piracy, the streaming transition itself , but each of those shifts changed how existing music was accessed. The current wave is different because it changes who, or what, is making the music in the first place. Watching how Deezer's disclosure translates into actual policy changes, and whether it prompts the larger platforms to follow suit with data of their own, will be the story to track over the months ahead.
Also read: AI research is increasingly optimized for conference acceptance and the field may be paying a hidden scientific price • The AGI race has no finish line and everyone is arguing about where it starts • Viral TikTok skits are exposing how confidently ChatGPT, Gemini and Grok get basic facts wrong
