I love making music with AI tools, and I love listening to music made by people. Over the last couple of weeks, the three streaming services I hear about most have each said something meaningful about how they will tell those two apart, and I think it is worth slowing down to look at what they actually announced.
Spotify is turning down the volume on slop
On October 20, Spotify shared an update on how it is handling AI-generated content. The figures, all from Spotify:
- More than 50 million slop and spam tracks taken out of recommendations this year.
- More than 75 million spammy tracks removed from the platform altogether for serious or repeated violations of its rules.
- Tracks with indicators showing a high probability of being AI-generated are 10 times more likely to fall below its 1,000-stream royalty threshold, which has been in place since 2024.
That last point is the clever one. Spotify describes the threshold as a first line of defense for human artists, because it makes it harder for anyone generating music at scale to siphon royalties away from real musicians. Spotify also says it now labels thousands of AI Personas, meaning artist identities it believes are AI-generated.
YouTube Music puts a label on the player
On October 14, YouTube announced that YouTube Music is rolling out labels for songs “created primarily using generative AI.” YouTube gives an example: using a generative model to create the composition and vocals. When AI played a significant role, you see a label right on the player page, and tapping it gives you more detail.
How does YouTube know? It says it is teaming up with music partners and using its own detection tools, applying the same labeling principles it already uses across YouTube. It also says these music labels will expand to the main YouTube app, which matters, because that is where a huge amount of casual listening happens.
Deezer hands the switch to you
On October 13, Deezer announced an on/off toggle that lets you exclude fully AI-generated tracks, albums and related artists from search results. Deezer says it will arrive globally before the end of 2026, so it is a promise for now. The company also says it receives nearly 90,000 fully AI-generated tracks every day, more than half of its daily deliveries on peak days, and that it has detected and pulled more than 42 million AI-generated tracks from recommendations between January 2025 and August 2026. Those are Deezer’s own numbers.
What I like about Deezer’s approach is that it treats listeners as adults. If you want to hear AI music, you can. If you would rather focus on human creators, you flip a switch.
Meanwhile, AI keeps moving into everyday life
These music changes are arriving while more people bring AI into their daily routines, and with a healthy dose of caution about it.
SurveyMonkey’s quarterly AI sentiment study, published October 15, surveyed 18,853 US adults between July 1 and September 30. It found that 40% use AI daily or weekly, up from 31% a year earlier. Only 18% said AI in a product makes them want that product more, roughly level with 16% a year ago. So adoption is climbing while the “AI inside” sticker still does very little to win people over.
The Allianz Pulse 2026 surveyed about 1,000 adults in each of eight countries: Germany, France, Italy, Spain, Poland, Austria, the UK and the US. Allianz reports that 48% use AI at least weekly, with a striking generational split of 77% among Gen Z and 23% among Boomers. Among AI users, 53% use it mainly for private purposes, which tells me AI is showing up first in everyday life, in the same places we listen to music. Allianz also classes 52% of respondents as AI “detractors”, scoring their overall view of AI between 0 and 6 out of 10, against 18% “promoters” scoring 9 or 10.
Put those together and I see a clear thread. People are happy to use AI as a tool in their own hands, and they want to know when it is shaping what they consume. Labels and filters respond to exactly that.
The honest limits
I am genuinely glad these labels exist, and I also want to be clear about what they can and cannot do.
- Detectors only hear the finished audio. They cannot see the creative process, so a track where a human wrote every note and used AI for one layer may be judged very differently from what its maker would describe.
- Definitions differ. YouTube talks about songs “created primarily” with generative AI. Deezer’s toggle targets “fully” AI-generated music. Spotify’s labels focus on AI artist identities. The same song could be treated differently on each service.
- Platform numbers are self-reported. Spotify’s and Deezer’s figures come from the companies themselves, and their detection methods are not independently audited in what they published.
- Surveys measure different things. SurveyMonkey and Allianz both run web panels, and they define “use” differently: daily or weekly in one, at least weekly in the other, across different countries. I would treat them as signals pointing the same way, without lining the numbers up side by side.
What I would love to see next
If you make music with AI, I would encourage you to disclose it proudly and early. Labels work best when creators supply them, and honest credit builds trust with listeners far faster than any detector can. If you are a listener, try the labels and toggles as they land and notice how they change the way you discover music. And if you build products, take the lesson from those survey numbers: people welcome AI that helps them, and they want to stay in control of it.
Read the sources
- Spotify: Prioritizing human artistry
- YouTube: YouTube Music adds new labels for AI music
- SurveyMonkey: AI Sentiment Study
- Allianz: Allianz Pulse 2026 on AI