Spotify's best AI-slop detector turned out to be the skip button
SummarySpotify says it pulled more than 50 million AI 'slop and spam' tracks out of recommendations this year, after finding that the songs listeners skipped, ignored or rejected were consistently the slop.
· MODEL Editors · 2 min read
Some links may be affiliate links. Disclosure

Spotify spent a year building defenses against AI-generated music. One of its clearest signals turned out to be the oldest one in streaming: people hitting skip.
What Spotify said. In an October 20, 2026 update, Spotify said it has taken more than 50 million "slop and spam" tracks out of recommendations so far this year. That's on top of the "more than 75 million spammy tracks" it says it has removed from the platform altogether for serious or repeated rule violations.
How it found them. Earlier this year Spotify looked at the tracks listeners were "skipping, ignoring, or telling us they didn't want to hear," and says "a clear pattern emerged: these tracks were consistently ones identified as slop or spam." Over the past six months it updated its recommendation system to push those down and surface more human-made music, and its editors don't program tracks flagged as slop. Spotify says listener complaints about slop in recommendations have since fallen markedly.
The detectors are just one input. Spotify says no single signal catches slop on its own. It combines listener engagement and feedback, upload patterns, metadata and "multiple third-party AI detection tools," and admits that "AI detection technology in particular has limitations, and it will only get harder as generative AI evolves." It also hunts for mass uploads, duplicates, SEO tricks, spammy playlist networks and artificially short tracks.
The other tools. More than 300,000 artists, covering over 99% of what listeners search for, now carry a "Verified by Spotify" badge. Thousands of AI-generated artist identities are labelled "AI Persona" and left out of recommendations unless you follow them. Artists submit tens of thousands of "AI Credits" a day disclosing where AI was used. And Spotify says tracks that look likely to be AI-generated are 10 times more likely to fall under its 1,000-stream royalty threshold.
The catch. "Out of recommendations" isn't the same as gone. As Engadget points out, the 50 million tracks may still be on the service, and the 75 million removals figure is the same one Spotify gave in September 2025, with no update.
Our take
For anyone building AI detectors, this is a humbling data point: at Spotify's scale, a huge pile of human "no thanks" signals was among the most useful ways to find slop, with the classifiers backing it up. For people making music with AI tools, the line Spotify is drawing is disclosure and quality, not AI itself. Credit the AI and make something people don't skip.
What we did and didn't do: figures come from Spotify's October 20, 2026 newsroom post, with coverage from NBC News, Billboard and Engadget, read October 22, 2026. We didn't test Spotify's systems, and Spotify hasn't said how many of the 50 million tracks were AI-generated rather than other spam. Not sponsored; no affiliate links.
MODEL Editors
Sources
- One Year In: Putting Human Artistry First on Spotify (Oct 20, 2026)
Spotify
- Spotify says it removed 50 million 'slop' and spam AI tracks this year (Oct 20, 2026)
NBC News
- Spotify Is Dancing Around Its AI 'Slop And Spam' Problem (Oct 21, 2026)
Engadget
- Spotify Shares Update on 'AI Spam,' Says 50M Songs Have Been Removed From Recommendations (Oct 20, 2026)
Billboard
FREE · EVERY WEEK
The top AI tools this week, tested.
Every week MODEL's editors pick the AI tools worth your time, try them hands-on, and tell you what's good, what's not, and who it's for. Free, in your inbox.
Free. Unsubscribe anytime. Our picks are editorial and can't be bought.
The weekly three is editorial. A subscription does not buy a place in it.
Continue

AI News
Americans rate AI's odds of easing their chores about the same as its odds of ending humanity
A new ABC News/Washington Post/Ipsos poll finds 57% of Americans say AI is bad for society, two-thirds oppose new data centers, and as many see a big chance AI makes everyday tasks easier (34%) as see a high risk of human extinction (33%).
· MODEL Editors

AI News
OpenAI's models are coming to Google Cloud. Microsoft is still the primary partner.
GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are rolling out in private preview in Google Cloud's model garden, billed on a Google Cloud invoice. OpenAI is now on all three big US clouds. Prices, general availability and some security features are still to come.
· MODEL Editors

AI News
AI finally learned COBOL. Now it has opinions about rounding
OpenAI says its newest models are far better at COBOL, Fortran and Visual Basic, the old languages behind a lot of banking and government software, and published two tests to show it. The numbers are OpenAI's own.
· MODEL Editors