It's not hard to imagine a song climbing the Billboard Hot 100. What's harder to imagine is the person who sang it not being able to pull it off himself. In several recent appearances, Fenix Flexin has either failed to lip-sync in time or opened his mouth only to discover his own vocal range can't reach the original track, forcing DJs to quietly layer the studio vocals back in to save the performance — and even that trick got caught by listeners. That's the most embarrassing part about "Rubberz": a song that landed a national chart position, yet the man behind it can't recreate it on stage.
Fenix Flexin was originally part of the LA rap duo Shoreline Mafia, known for West Coast trap. This solo release, "Rubberz," takes a sharp left turn — heavily drawing from '80s British synth-pop, with a lead melody that clearly echoes Morrissey, right down to a deliberately affected British accent. The song currently sits at No. 58 on the Billboard Hot 100. Fenix himself has denied accusations that AI wrote the track, but he hasn't offered anything solid enough to shut down the doubters.
Just listening to the recording quality raises red flags. The hi-hats have a thin, brittle digital quality, the chorus vocals sound like a low-bitrate MP3, and the reverb tails often cut off abruptly before finishing naturally. Switched on Pop host and songwriter Charlie Harding described it this way: "There are these ghostly harmonies that phase in and out for no reason, and the drum reverb is full of unnatural digital artifacts, like it was pulled from a 64kbps MP3 off Napster." He believes this is because most generative AI music companies train their models on unlicensed, low-quality MP3 source material, which produces this degraded, artificial instrument tone.
Running the full track through AI detection tools didn't yield anything conclusive — I tested five different tools, and they placed the probability of AI or AI-assisted generation somewhere between 20% and 30%, and these tools themselves aren't especially reliable to begin with. But the surrounding materials weren't so lucky: the single's cover art and a post celebrating the chart debut were flagged by multiple AI image detection tools with over 97% confidence as AI-generated. Running the lyrics through AI writing detectors and Gemini also returned results suggesting AI authorship. YouTuber and producer Curtiss King ran the lyrics through Claude and got multiple signals pointing to AI generation as well.
The lyrics themselves read like logically coherent but hollow stacks of words. The rhymes are cleaner than you'd expect from a rapper with real chops — no slant rhymes, no internal rhymes, everything strictly follows an AA-BB structure, as if the rhyme scheme itself mattered more than the narrative: "Swipin' cards and stackin' chips, I saw you sinkin' ships / Left me standin' in the pourin' rain, now I bought a heavy diamond chain." Who's sinking whose ships? The image just appears and then vanishes without a trace. In the chorus — "Bankrolls on the table / Watchin' local cable / Hatin' that I'm able / Hatin' that I'm able" — the lyrics never once explain what exactly Fenix is "able" to do, or why anyone would hate him for it.
Cracks That Live Lip-Syncing Can't Paper Over
In the third verse, the line "Not the jewelry" cuts off as if the sentence was chopped mid-thought. Harding described the whole track as having "absolutely no flow," saying the delivery of the title line, "I just let the rubber band snap," is so awkward that "almost no competent rapper would phrase it this way — the syllabic stresses land in all the wrong places."
What's even more troubling is how he looks on stage. In his earlier studio appearances, Fenix was visibly struggling through an awkward lip-sync, as if he hadn't even memorized his own lyrics. And when he did actually try to sing live, the problems became even more obvious — multiple videos show him completely unable to replicate the vocal range and British accent from the recording, forcing show producers to mix the original studio vocals in at a very low volume underneath his live performance. In some segments, the DJ even deliberately cut the backing track entirely to "prove" it was a real live vocal — which only made the gap even more glaring.
Facing the backlash, Fenix's explanation is that the studio version used AutoTune and similar vocal processing, which accounts for the big difference from his live performances. But Harding, Curtiss King, and music critic Anthony Fantano aren't buying it. King compared AutoTune to spell-check — even when pushed to its limits to create a distinctive robotic tone, its purpose is to correct mistakes, not to extend a range someone doesn't naturally have or conjure an accent out of thin air. Fantano holds the same view.
Ironically, the very badness of the song has become the strongest argument that it might actually be human-made. Harding put it this way: "I would've assumed this was written by someone who'd never touched an '80s kids' toy keyboard in their life." If "Rubberz" really was AI-generated, it clearly used an inferior model — and he admits what worries him is that listeners are finding something interesting in a piece made with a subpar model from some second- or third-tier generative AI music company, "when more advanced models have already made huge strides in smoothing out the obvious noise and vocal weirdness."
Fenix later posted a few screenshots that appear to be from a Pro Tools project file, but that doesn't really prove anything. Unless he releases the raw recordings and the full project files, the AI-ghostwriting suspicions probably aren't going away anytime soon.
