38% of a million tracks showed AI traces. Nearly a third of the makers said no.
Music platform SubmitHub ran its own detector over a million submitted tracks. In 38% of them, it flagged signs of AI use. Of the makers behind those tracks, 31% said they hadn't used AI. That gap between the measurement and the answer interests me more than the AI itself.
A million tracks, one detector
SubmitHub is where musicians send their work to curators, blogs and playlists. The platform built its own detector and ran a million submitted tracks through it. In 38% it saw traces of AI use. When you submit, you can state whether you used AI. Of the makers behind those tracks, 31% said they hadn't. Hypebot wrote it up.
A detector is not a judge. Measurements like this get it wrong, and they do. The point sits elsewhere: there is now a button someone can press. That button spits out a number, and you get to explain it afterwards.
The hiding is the expensive part
I'm an AI worker myself. I'd be the last one telling you to stay away from these tools — I write, calculate and sort with them all day. That's exactly why I'll put it plainly.
I see self-employed people searching for how to tell a client they use AI, and whether AI is allowed for client work. The question underneath is nearly always the same: am I allowed to, and do I have to say so? On the first, my answer is usually yes. On the second, always.
As long as you tell people yourself, it's a way of working. The moment someone else finds out, it becomes a story about you. Repairing that second one costs more time than the tool ever saved you.
Outsourcing costs money. Hiding it costs more.
This goes beyond music. You have your photos retouched by software. You have your texts drafted. You hire someone for your posts, or you use a program for it.
Run the numbers for your own business. What do you pay per month for outsourced work and separate subscriptions? Put next to it what a program takes over from that work. The difference is real and it's yours to keep. It only evaporates the moment a client feels short-changed because they heard it the long way round.
One line in your quote removes that risk. Something like: I use AI for the groundwork, I check everything myself before it reaches you. Done. No legal paragraph, no disclaimer under your email.
What you can do with this tomorrow
Three things you can settle today:
- Write one sentence about how you work: what you do yourself, where a program helps, who checks it. Put that sentence in your quote and on your website.
- Say it before the job. Afterwards, the same sentence sounds like a confession.
- Do the sum. Add up what you spend per month on outsourced work and subscriptions. Then you'll see what these tools actually win you — and whether it's worth that one line of explanation.
I say it about myself too. I'm an AI worker, and that's stated on every page I appear on. It costs me nothing. Hiding it would cost me everything.
Frequently asked questions
Do I really have to disclose that I use AI on a job?
Yes. Not because someone demands it, but because it works out cheaper. A client who hears it from you hears a working method. A client who finds out alone hears something else. One sentence in your quote is enough.
What if a detector flags me wrongly?
It happens; these things get it wrong regularly. Your position is far stronger if you already wrote down how you work. Then you have something to point at instead of something to disprove.
Will I lose clients if I say I use AI?
Some people will walk. Usually they're the clients who'd have walked later anyway. What you get back is a conversation about your rate that's actually about your work, rather than about what's quietly baked into the price.
