Re:AI is creating a 'human premium' for art created by people
by tolugenius on Sunday October 11, 2026 @07:59AM
(#50042318)
Curious on two things
1. How often do you encounter false positives, and does that change things?
2. Have you considered outright prohibiting/banning vs disclosure?
Just want to hear from someone running a platform how they deal with this since I imagine it's very hard to get right.
Re:AI is creating a 'human premium' for art created by people
by jasongrishkoff on Tuesday October 13, 2026 @12:08PM
(#50073982)
1. Rare. 99.4% accurate.
2. Yes, in September we outright banned slop. Got a lot of backlash for that, but I think it was for the better.
Re:AI is creating a 'human premium' for art created by people
by gchamonlive on Sunday October 11, 2026 @07:55AM
(#50042291)
How do you handle false positives in the detector?
Re:AI is creating a 'human premium' for art created by people
by jasongrishkoff on Tuesday October 13, 2026 @12:07PM
(#50073952)
We have an appeal process. FP rate is super low - accuracy around 99.4%.
Re:AI is creating a 'human premium' for art created by people
by andsoitis on Sunday October 11, 2026 @07:57AM
(#50042308)
What’s your site?
Re:AI is creating a 'human premium' for art created by people
by jasongrishkoff on Tuesday October 13, 2026 @12:06PM
(#50073935)
SubmitHub
On my site, we started asking artists to disclose their use of AI starting ~2 years ago. Turns out roughly 1 in 3 lied.
So, I developed an AI detector. It's working pretty well for that. But it's a schlep having to keep up with all the new models. We're processing ~60,000 songs a day at this point, kinda crazy!