AI Music Tool Treblo Says Its Own Detector Found Its Fingerprints All Over Fenix Flexin's Hit 'Rubberz'
The company built a classifier to spot songs made with its technology. The first song it flagged with high confidence may be on the Billboard Hot 100.

Key points
- Treblo released an open-source AI music classifier on Monday that detects songs generated by its own tool, with a false positive rate below 1 in 1,000.
- The classifier rated Fenix Flexin's song "Rubberz" as "very likely Treblo" with high confidence.
- Treblo CEO Ryan Tremblay told The Verge the song would be "the first known AI-generated song to reach the Billboard Hot 100" if confirmed.
- Fenix Flexin has denied using AI and posted what he claims are original session files, but critics say the clips are unconvincing.
A song either charted on the Billboard Hot 100 with the help of an AI tool, or it did not. The answer now has a company, a number, and a press statement behind it.
On Monday, Treblo, a company that makes an AI music generation tool, released something called the Treblo AI Music Classifier. A classifier, in plain terms, is a piece of software trained to recognise patterns. This one was trained to spot songs that Treblo's own tool produced. The company says it gives a false answer fewer than once in every 1,000 tests.
The first notable song it examined was "Rubberz" by rapper Fenix Flexin.
What did the classifier actually find?
The software returned a high probability score, and Treblo says the result points in one direction. "The classifier returns a very high probability score for Rubberz, indicating the released audio was generated almost entirely by our model," CEO Ryan Tremblay said in a statement to The Verge AI. He added: "Someone appears to have used our model to find a sound that millions of people connected with and that's exciting."
That second sentence is doing a lot of work. Treblo is not playing the injured party here. The company is, by its own admission, enjoying the attention.
Does Fenix Flexin accept any of this?
No. Fenix has denied the claims and posted short video clips he says show the original recording session project files. Those videos have not quieted the allegations. Musician Medasin, who first raised suspicions publicly, argued in an Instagram story (since expired, first reported by The Verge) that Fenix simply used an AI stem separator, a tool that splits a finished track into separate parts, to fabricate the look of a real session.
Fenix did not respond to a request for comment before publication.
What does this mean for music fans and working musicians?
For listeners, the short answer is: a song you may have genuinely liked might have been made without a human in the room. That is not a legal problem yet, but it is a transparency problem the industry has not solved.
For working musicians, the harder message is that detection tools now exist and companies will use them. Treblo's classifier only catches Treblo-made tracks, not music from other AI tools. Still, the direction is clear: AI-generated music is becoming traceable, at least in some cases.
One honest caveat worth naming: a less-than-1-in-1,000 false positive rate sounds tight, but it means a small number of real, human-made songs could be flagged incorrectly. That is worth keeping in mind before anyone draws a final conclusion.
Takeaway: If you make music and wonder whether AI-detection tools could affect your work, it is worth knowing they exist, what they can and cannot catch, and that the first Billboard Hot 100 test case is already playing out in public.



