Back to discover
NewsLearn & research

YouTube Music starts labeling tracks made primarily with AI

An October 14 announcement introduces player labels for music made primarily with generative AI; expansion to the main YouTube app is still ahead.

Conceptual illustration of a vinyl record dissolving into purple particles and musical notes, with a blank amber tag.
CurioFuse · AI-generated illustration

YouTube says its new music labels identify tracks where AI contributed substantially, including generated compositions and vocals. Listeners can tap the label on the player page for further information. The company combines information from music partners with its own detection tools to identify such recordings.

The partner guidance distinguishes wholly generated material, partial use and no generative AI. An omitted declaration remains unknown unless other signals establish a classification. A generated musical layer combined with live performance is a partial-use example; ordinary pitch correction and automated mastering appear among examples without generative AI. These categories describe contributions, rather than treating every software-assisted recording alike.

The earlier video rules provide narrower background: automatic labeling applies when a creator leaves AI use unspecified and significant photorealistic AI is detected. Creators can correct disclosure in Studio, but permanent exceptions include YouTube’s own AI tools and C2PA provenance metadata indicating fully generated content. Those conditions concern the video policy.

Ars Technica’s commentary highlights the listening problem: people may need to check their phone or background tab to notice a disclosure. It also warns against treating an absent label as evidence of exclusively human creation.

Sources(4)

Sources & standardsReport a correction