YouTube Music has begun placing a label on the player page when a track was created primarily with generative AI. YouTube gives composition and vocals generated by a model as an example, and says the labels will also expand to the main YouTube app.[1]
The label is a provenance signal: it tells the listener something about how the audio was made. It is not a verdict on whether the track is lawful, original, good, misleading or worth recommending.
The metadata has more than two states
Music partners can declare AI use through DDEX ERN 4.3.2 or later, CSV delivery or YouTube Studio. YouTube’s current guidance recognises fully generated, partly generated and no generative-AI content. If no value is supplied and YouTube cannot decide from other signals, the state is unknown.[2]
What the four states actually say
| State | YouTube definition | What it does not settle |
|---|---|---|
| Fully Gen AI | All content was created using generative AI | Who directed the process or owns the rightsSource 2 |
| Partly Gen AI | Part of the content was created using generative AI | How large or artistically important that part wasSource 2 |
| No Gen AI | No content was created using generative AI | Whether non-generative digital tools were usedSource 2 |
| Unknown | No usable declaration or determination | That the track is human-madeSource 2 |
A visible label can still carry an invisible threshold
The public announcement says the player label appears where AI played a significant role and describes tracks primarily made with generative AI. The delivery metadata, however, can contain a partly generated designation. The user-facing experience therefore depends on how YouTube maps a granular declaration and its own detection signals into the label that listeners see.[1][2]
That mapping matters. YouTube’s own guidance puts a generated instrumental layer, AI-assisted lyric writing and human-performed audio accompanied by an AI-generated video under partly Gen AI. Ordinary pitch correction and automated mastering are listed under no Gen AI. The declaration concerns a particular creative use of generative tools, not whether any software marketed as AI touched the work.[2]
Detection and disclosure share the job
YouTube says it is working with music partners and using its own detection tools. Its broader labelling system can also use internal signals and C2PA metadata, while allowing creator corrections in many cases. YouTube has separately said that an AI label alone does not determine recommendation treatment or monetisation eligibility.[1][3]
This hybrid approach is understandable. Self-disclosure has information that an automated detector may never infer reliably, especially when AI is one step in a long production chain. Detection can catch omissions, but it also creates false-positive and appeal problems. Neither route is sufficient on its own.
A useful transparency system should make the first answer easy without encouraging people to assume the other answers. YouTube Music’s label is a meaningful start precisely because it is limited. The next test is whether the detail behind the badge becomes specific enough to support the different decisions listeners and creators actually need to make.
Sources & further reading
- YouTube Music adds new labels for AI music YouTube · 14 Oct 2026 · Accessed 15 Oct 2026
- Disclose Gen AI usage for music content YouTube Help · Accessed 15 Oct 2026
- Improving AI labels for viewers and creators YouTube · 27 May 2026 · Accessed 15 Oct 2026