According to The Verge, the music streaming service Deezer is opening up its proprietary AI-generated music detection tool for other platforms to use. CEO Alexis Lanternier stated that a majority of AI music uploaded to Deezer is done so with fraudulent intent, and the company demonetizes every fraudulent stream it finds to protect royalties for human artists and rights holders. This announcement follows actions by competitors like Spotify, which began rolling out new AI and impersonation policies last year and is developing a new metadata standard for AI disclosure. Meanwhile, Bandcamp is taking the hardest line by completely banning AI-generated content. The core issue is that AI music creation tools from companies like Suno and Udio are producing tracks that are increasingly difficult to identify as synthetic.
The Detection Arms Race
So, how do you even detect this stuff? It’s not like there’s a watermark or a signature you can just scan for. Deezer’s tool, which they’ve been using internally, likely analyzes a bunch of sonic and metadata fingerprints that human-made music typically has—or that AI music typically lacks. Think about patterns in composition, mixing, or even the way files are uploaded and tagged. But here’s the thing: this is a moving target. As the AI models from Suno and Udio get better, they’ll start mimicking those human patterns more closely. It’s an arms race. The detector has to constantly learn, just like the generators do. That’s probably why Deezer is opening it up; more data from more platforms makes the tool smarter for everyone involved.
Why This Really Matters
Look, it’s not really about being anti-AI. It’s about fraud and the economics of streaming. Lanternier’s quote is key: they see AI music primarily as a tool for “commit fraud.” What does that mean? Basically, bad actors could use AI to flood a service with thousands of tracks, then use bot farms to generate fake streams and siphon royalty money out of the pool meant for actual artists. That’s the immediate, tangible threat. The longer-term philosophical debate about AI art is secondary to stopping that drain. Demonetizing these tracks is the direct financial fix. But the broader industry push for labeling, from Spotify’s metadata plans to Deezer’s detector, is about transparency for listeners. Do people want to know if they’re listening to a machine? Probably. Should they have that choice? I think so.
A Fractured Industry Response
What’s fascinating is the complete spectrum of responses we’re seeing. You’ve got Bandcamp on one extreme, just saying “no” entirely—which fits their artist-centric, indie ethos. Then you have Deezer in the middle, trying to build a technical filter and share it. And then there’s Spotify, focusing more on disclosure and policy. There’s no consensus. And why would there be? The technology is evolving weekly, and the legal landscape is still a fog. This patchwork approach is what happens when an industry gets hit with a disruptive wave it didn’t see coming. The big question is whether any technical solution can ever be perfect. Or will it just become a constant game of whack-a-mole, where the fraudsters are always one step ahead? For now, tools like Deezer’s are the best defense the streaming economy has.
