Study Finds Nearly 40 Percent of Music Released in July Involved AI

Here's a stat that should make you sit up: almost four out of every ten tracks released in July had AI somewhere in their DNA. That's the headline finding from a new SubmitHub study that analyzed over a million releases from last month using SH Labs, the AI-detection tool the music submission platform launched earlier this summer.

The breakdown is even more interesting than the topline number. According to the study, 38.5 percent of tracks showed some form of AI involvement — 23.2 percent were entirely AI-generated, while another 15.3 percent featured AI-generated audio that humans had modified or processed afterward. So it's not just a flood of fully synthetic songs; there's a growing gray zone where producers are blending machine output with their own work.

And here's the part that stings. When SubmitHub reached out to artists whose tracks got flagged, 31 percent flat-out denied using AI tools. Whether that's embarrassment, fear of being demonetized, or just people not counting AI stem generators as “using AI,” the honesty gap is real — and it's exactly why detection tools like this exist in the first place.

“The path forward for AI music is disclosure,” SubmitHub founder Jason Grishkoff said. “People should be able to decide for themselves whether they want to engage with AI-generated music.” Hard to argue with that. Nobody's putting this genie back in the bottle, but knowing what you're listening to feels like a pretty low bar to clear.

The numbers track with what other platforms have been reporting all year. Back in May, Deezer said AI-generated music made up 44 percent of its daily uploads — roughly 75,000 tracks a day. Since then, the industry response has snowballed: Deezer demonetized the vast majority of AI music streams, Spotify purged tens of millions of spam tracks, and Bandcamp banned AI music outright. Dance music's own gatekeepers are moving too — just this week, Beatport announced it's banning fully AI-generated tracks and tagging AI-assisted ones so DJs know what they're buying.

For electronic music specifically, this all cuts close to the bone. Ours is a genre built on machines, samples, and synthesis, so the line between “tool” and “ghost producer that never sleeps” is blurrier here than anywhere else. Is a producer who generates stems and arranges them by hand doing anything different from someone flipping a sample pack? That's the debate the scene is going to be having for a while.

One caveat worth keeping in mind: the study acknowledges the disclosure problem cuts both ways, and detection isn't perfect. But even with a margin of error, the direction is obvious. AI music isn't coming — it's already here, and it's charting a bigger share of the release pile every month.