SubmitHub now blocks any track its detector scores 85% likely AI
My stack My fav 6 tools for AI music after testing 70+
-
ElevenMusic · Music generation -
Moises · Stem separation -
Controlla · Voice swap & cloning -
Cryo Mix · Mixing + mastering -
Neural Frames · Music videos -
ONCE · Distribution
Some links are affiliate links — I earn a commission at no cost to you. Every tool here is one I run myself and truly believe in. This stack is proven.
A month ago SubmitHub published the number: 23.2% of the songs coming through its pipeline were fully AI-generated. On September 17, 2026 it did something about it. “Our new policy: we’re blocking pure AI submissions,” founder Jason Grishkoff wrote in an update on the site, first flagged by Music Ally.
Where SubmitHub’s 85% line falls
The cutoff is a detector score, not a declaration. Anything SubmitHub’s own detector flags at 85% or higher likelihood of being AI gets blocked.
Grishkoff says he tested the line at 75%, 90% and 95% before landing on 85%. The score itself is a blend of two readings, pure and hybrid, weighted roughly 3 to 1 toward the spectral result, which is the detector’s analysis of the audio itself rather than the metadata around it.
Hybrid tracks get caught too. A hybrid score above 85% usually means a track was generated on a tool like Suno and then given a few small tweaks after download.
There is no appeal and no workaround. Writing the lyrics or the composition yourself does not exempt a track, and curators do not have to change any settings.
Why Grishkoff points at Spotify rather than at AI
This is the part worth reading properly, because the argument is commercial rather than moral.
Spotify spent most of 2026 staying out of the fight. In February its co-CEO Gustav Söderström said the platform “should not decide what kind of tools you’re allowed to use,” comparing AI to an electric guitar or a DAW.
Then in August Spotify announced an AI Persona label attached to artist profiles, with AI artists kept out of recommendations by default. Grishkoff’s read is that a track his detector scores at 85% is probably getting flagged on Spotify too.
Because our job at SubmitHub is to create connections and ultimately help artists get heard, it feels wrong for us to suggest that AI artists use our product if it's unlikely to work.
What the policy costs SubmitHub
Roughly 11% of the tracks uploaded in the past month scored 85% or higher, and Grishkoff puts the revenue hit at about 10%.
“We’re actually ‘cutting off’ 10% of our revenue stream here, so this decision goes counter to traditional business logic,” he wrote.
He is also honest that it will misfire sometimes. He cites a third-party measurement of 99.4% accuracy and expects the rollout to be messy at first, with the 85% number the most likely thing to move.
Hot or Not and Meta Ads stay open to AI creators, though he adds that ads are just as futile against Spotify’s changes.
SubmitHub joins a short list of platforms with a line
Deezer built its own detector, started flagging AI tracks and pulling them out of recommendations and editorial playlists. Tidal did something similar, and Bandcamp banned AI outright.
What makes the SubmitHub move different is that it sits at the promotion layer rather than the listening layer. A streaming service decides what gets recommended. A promo platform decides whether you can pitch at all.
If you make music on a generator and post-process it, the number that now matters to your release plan is your own detector score, not your opinion about where the line should be. Grishkoff sampled four cutoffs before picking one, and 85% is where he stopped.
Frequently asked questions
What is SubmitHub's new AI policy?
SubmitHub is blocking submissions that its in-house detector scores at 85% or higher likelihood of being AI-generated. Founder Jason Grishkoff announced it in an update on the site, describing it as blocking pure AI submissions. Tracks scoring below that line are still accepted.
What AI score gets a track blocked on SubmitHub?
85% or higher. The figure blends the detector's pure and hybrid readings, weighted roughly 3 to 1 toward the spectral result. Grishkoff says he tested cutoffs at 75%, 90% and 95% before settling on 85%.
Can you still use SubmitHub if you used AI to help write a song?
Yes, as long as the track scores under 85%. Grishkoff says a hybrid score below that line signals that humans were involved in the process. A hybrid score above it usually means the track was made on a generator and then tweaked slightly after download.
Do SubmitHub's Hot or Not and Meta Ads still accept AI music?
Yes. Grishkoff says those two products sit outside the new rules. He adds that running Meta Ads is still unlikely to help much given Spotify's recent changes.
How accurate is SubmitHub's AI detector?
Grishkoff says a third party measured it at 99.4% accurate. He acknowledges the policy will get some calls wrong, and argues that a track his detector flags at 85% or higher is probably being flagged by Spotify too.
About the author
Covering AI music tools, industry shifts, and news for music creators and professionals. Twice-weekly newsletter at aimusicpreneur.com.
