YouTube Music has started labelling songs that were made with generative AI. The announcement, on 14 October, comes as streaming services struggle with a flood of machine-made tracks. On the same day, Deezer said it is building a filter that will let listeners hide AI music altogether.
Here is how the YouTube labels work, what counts as “AI-made”, and what neither company has said yet.
What listeners will see
When AI played a significant role in a track, the label appears on the song’s player page. Tapping it opens a menu with more detail. YouTube’s example is a song where a generative model was used for both the composition and the vocals. The labels will also reach the main YouTube app.
YouTube says it identifies these songs by working with music partners and by using its own detection tools. The announcement does not say which countries get the labels first, or on what timetable, so we cannot tell UK listeners when to expect them.
How a song gets labelled
Labels depend heavily on what music partners, such as labels and distributors, tell YouTube. Partners declare AI use with one of three values: “Fully Gen AI”, “Partly Gen AI” or “No Gen AI”. If a partner gives no answer, YouTube may use other signals to decide.
YouTube’s help page gives examples of each category:
- Fully Gen AI: a track generated from a text prompt, or a song whose composition, instrumentals and vocals are all made by AI.
- Partly Gen AI: AI-made layers with a human performance on top, or lyrics or melodies written with AI help.
- No Gen AI: standard pitch correction and automated AI mastering.
That last line matters. A song tuned with pitch-correction software does not count as AI-made.
Disclosure, not a ban
The labels tell you something about a song; they do not stop it being played or paid. According to TechCrunch, AI involvement alone does not make a track ineligible for monetisation on YouTube. The company is also testing Dream Track, which lets selected creators generate soundtracks with AI versions of the voices of participating artists.
Why platforms are acting now
Deezer says that in July it was receiving nearly 90,000 fully AI-generated tracks a day, which was more than half of its daily uploads. It said about 10% of uploads were AI-generated when it launched its detection tool in January 2025. These figures come from Deezer through TechCrunch.
Deezer’s new filter will be a toggle that hides AI-generated tracks, albums and related artists from search results. It is still in development, and Deezer expects it before the end of 2026. A Deezer-Ipsos survey from November 2025 found that roughly 45% of its customers wanted a way to remove AI music from what they hear.
Other services have gone their own way, as TechCrunch summarises:
- Spotify labels AI tracks and has a spam filter.
- TIDAL labels recordings it finds to be entirely AI-generated, lets listeners turn them off and has cut off their monetisation.
- Bandcamp bans AI tracks.
- Apple Music is reported to be planning “Made With AI” labels by the end of the year.
YouTube’s approach, in short, is to disclose rather than restrict.
What we do not know
Two gaps stand out. YouTube has not said where the labels are available. And a label only appears when a song is declared or detected, so an unlabelled track is not proof that no AI was used.
We have not tested the feature ourselves, so we cannot say how often the labels appear or how accurate they are. We will update this article when YouTube gives a timetable and when Deezer’s filter goes live.
Leave a Reply