YouTube required creators to disclose realistic synthetic media

YouTube's blog and help centre document a disclosure requirement for realistic AI content, explicitly listing AI-generated music among it.

Historical event

March 18, 2024

First source published

March 18, 2024

Site publication

September 18, 2026

What happened

On 18 March 2024, YouTube published ‘How we're helping creators disclose altered or synthetic content’, introducing a Creator Studio tool requiring disclosure when “realistic content” is made with altered or synthetic media, meaning content a viewer “could easily mistake for a real person, place, scene, or event.” The post states labels would roll out across YouTube surfaces “in the weeks ahead,” starting on mobile, and describes the requirement as building on a responsible-AI approach YouTube had announced the preceding November.

What the documents say

YouTube's living help-centre article, ‘Disclosing use of GenAI content’, current as retrieved on 16 September 2026, sets out the disclosure mechanism in more detail than the original post: creators select ‘Yes’ or ‘No’ under an ‘AI use’ attribute when uploading, and YouTube may also apply a label automatically for content made with YouTube's own generative tools, content carrying C2PA provenance metadata, or content its internal systems detect as AI-generated. The help page's list of examples requiring disclosure explicitly names “AI generated music,” alongside altered footage of real events and realistic fabricated scenes; it separately lists exemptions, including using a generator for scripts, captions, or one's own cloned voice for dubbing. The page states disclosure does not affect a video's monetisation eligibility or audience reach, but that consistent non-disclosure can lead to a manually applied label or removal from the YouTube Partner Program.

Why it matters for makers

The operative distinction is realism, not method: YouTube's own examples state that unrealistic or clearly stylised content, and edits that are “primarily aesthetic,” do not require disclosure, while realistic AI-generated music is named directly as something that does. This means a soundtrack or vocal track's disclosure obligation on YouTube turns on whether a viewer could mistake it for something real, a different test from Spotify's or Deezer's, which focus on stream authenticity or a “fully AI generated” content flag rather than realism.

What to check before you use it

This is an editorial checklist. Read the current help-centre examples list before uploading, since YouTube states the list is illustrative and continues to be updated. If your music was made or meaningfully altered with a generator in a way a listener could take as a real performance, select ‘Yes’ under AI use even if the visual content is unrelated. Check whether any automatic label has already been applied via C2PA metadata carried in your file, since the help page states some automatically applied labels cannot be adjusted by the creator afterward.

• Could a viewer mistake your AI-assisted audio for an unaltered human performance or real event?
• Does your upload carry C2PA or other provenance metadata that might trigger an automatic label regardless of your own disclosure choice?
• Have you checked the current help-centre examples list rather than relying on this article's summary of it?

YouTube's policy is unusual among the platforms covered here for naming AI-generated music as a specific disclosure example rather than leaving music to a general synthetic-media category, which makes its realism test a useful reference point even for a maker who releases primarily elsewhere.

Sources & reading trail

Announces the Creator Studio disclosure requirement, its realism test, and the rollout timeline.

Source published: 18 March 2024 · Retrieved: 16 September 2026

Living document giving the current mechanics, naming AI-generated music as an example requiring disclosure, and describing automatic detection and enforcement.

Source published: Not established · Retrieved: 16 September 2026

Papers, terms and official documents establish the record; the maker reading and the checks are Signal to Song editorial analysis. This retrospective draft does not imply the site published on the event date.

Continue reading

• A provenance standard records edits, not the truth of a file
• DeepMind watermarked its own generated music, not music generally
• Music’s AI credit standard is still being built by DDEX
• Browse the complete the archive

Published September 18, 2026, not on the date of the event described.