Seventy-five thousand. That's how many AI-generated songs Deezer says land on its platform every single day, roughly 44 percent of everything uploaded. Sit with that for a second. Nearly half of the new music arriving at a major streaming service was made by a machine, and a big chunk of it is junk built to siphon royalties from real musicians.
Deezer has been the loudest of the big platforms about this. Now it's taking the fight to other people's apps.
On June 11, the French company opened its AI music detector to anyone, on any platform. You don't even need a Deezer subscription. Point the tool at your Spotify, Apple Music, or YouTube Music account, and it tells you how much of your library was, allegedly, never touched by a human hand.
How the AI music detector actually works
The process is almost boringly simple, which is kind of the point. You go to Deezer's AI music detection site, pick your streaming service from a list of around 20 supported platforms (Spotify, Apple Music, SoundCloud, and YouTube Music all included), and grant access to your account.
From there, Deezer imports your playlists. According to The Verge, it appears to use Tune My Music for the job, the same tool Deezer already relies on when someone defects from a rival and wants their library to come with them. Once your playlists are in, the scanner runs, flags any tracks it thinks are synthetic, and hands you the results. You can share them, too, if you want to broadcast just how compromised your weekend mix really is.
Deezer claims its detection works with over 99 percent accuracy. That's a confident number, and it's worth keeping a little skepticism in your back pocket, because there's no easy way for a regular listener to check it. Then again, the company has wrestled with the flood of AI tracks more than almost anyone.
Why Deezer is doing the competition's homework
Here's the genuinely interesting part. Deezer didn't build this consumer tool because it wanted to. It built it because nobody else would license the technology.
The company was first among the majors to label AI-generated music. It then offered its detection tech to competitors, hoping the rest of the industry would buy in and clean up the mess together. That didn't happen. As Engadget reported, CEO Alexis Lanternier said no other company had followed Deezer's lead, so it decided to put the tool directly in users' hands instead.
The rest of the field went softer. Qobuz rolled out its own detection system. Apple and Spotify chose voluntary tagging, which means an AI track gets labeled only if whoever uploaded it bothers to be honest about it. You can guess how reliable that is when the whole game is deception.
So this release reads less like a feature launch and more like a competitive jab. Deezer is basically saying: fine, we'll show your users what you won't.
And the hook is shrewd. Lanternier noted that nearly half of the people switching to Deezer from another service arrive with AI tracks already lurking in their playlists. Translation: scan your Spotify library, get unsettled by the result, and Deezer is right there, ready to welcome you over. The detector is a public service and a recruitment pitch wearing the same coat.
The royalty theft hiding in your playlists
The detection tool would be a mere curiosity if AI music were harmless filler. It isn't, and that's the real story underneath the press release.
Deezer says that of the thousands of AI tracks uploaded daily, as many as 85 percent are fraudulent. We're not talking about a hobbyist generating an ambient soundscape for fun. We're talking about operations that clone the work of human artists, push the copies onto streaming platforms, and pocket royalty payments that should have gone to the people who actually made the original music.
It's stream fraud, dressed up in the language of innovation. And it scales horribly well, because spinning up another fake track costs almost nothing. A human songwriter spends weeks on a record. A bot farm spends seconds.
This is where Deezer's hardline stance starts to look less like marketing and more like principle. Where Apple and Spotify flag AI content, Deezer blocks it outright from its recommendations and editorial playlists. The synthetic stuff still lives on the platform, but it won't get promoted, won't surface in the lists Deezer's own editors curate, and won't quietly steal attention from working musicians. That's a real distinction, and frankly it's the line more services should be drawing.
What this says about streaming's bigger problem
Step back and the math gets bleak. If 44 percent of daily uploads to one major platform are AI-generated, the catalog available to listeners is being diluted at a pace no human industry can match. Every fake track competes for the same finite pool of listener time and royalty money as the real ones.
The voluntary-tagging approach favored by the biggest players essentially trusts the fraudsters to identify themselves. Lanternier expects the detector to be eye-opening, and he's probably right, if only because most people have no idea what's already in their own libraries. You add a song from an algorithmic recommendation, it sounds fine, you never think about who or what made it. The tool turns that invisible problem into a number you can actually see.
Whether it changes anyone's behavior is another question. Most listeners care whether a song sounds good, not where it came from. The detector might spark a moment of unease, then get forgotten by Tuesday.
What to watch next
The more meaningful test is whether Deezer's move shames the bigger platforms into action. Spotify and Apple have far larger user bases and far more sway over the upload pipeline. If a mid-sized French service can detect AI tracks with claimed 99 percent accuracy (that number again), the giants can't credibly argue it's impossible. They've simply decided tagging is enough.
There's a regulatory angle hovering in the background, too. As royalty fraud climbs and artists grow louder about it, the pressure to mandate disclosure, rather than leave it voluntary, will only build. Deezer just handed critics a working example of what's technically feasible.
For now, the AI music detector is a free tool you can point at your own playlists today. Run it once, if only out of curiosity. The result might tell you more about the state of streaming than any earnings call ever will.