YouTube AI disclosure: AI backdrop and own-voice clone examples

On YouTube's own list, cloning your own voice for voiceovers or dubs, and generating or extending a backdrop to simulate a moving car, are both edits creators do not need to disclose. Disclosing use of GenAI content puts them under minor edits and non-realistic content, and keeps disclosure for realistic content that is generated or meaningfully altered with AI (read 2026-10-02). The page says its lists are not exhaustive.

Voice cloning in Sume is an app feature, not an API one; the example matters when you finish a video with a cloned voice of your own.

Which edits does YouTube list as not needing disclosure?

The page splits the exemption into non-realistic content and minor edits, which it describes as primarily aesthetic and not altering the content in a way that could mislead the viewer about what actually happened.

Does a cloned voice need a label if it is not mine?

The page's example is cloning one's own voice. It also lists "making it appear as if someone gave advice that they did not actually give" among examples that need disclosure. A voice that is not yours, used in a way that implies the person said it, falls on the disclose side of the page's own lines. That reading is mine; the page does not name third-party voice clones explicitly.

What about a video with AI music under the voice?

AI generated music is on the list of content that needs disclosure. A video with your cloned voice, which is exempt, and a generated soundtrack, which is not, therefore needs the AI use answer set to Yes. YouTube asks the AI use question once per video, so one realistic AI element is enough.

What should I do?

Go through the audio and picture tracks separately. If any single element is realistic and meaningfully AI generated, answer Yes in the Attributes section under AI use. If all of them sit on the exempt list, No is the page's answer, and the page notes that YouTube may still label C2PA-bearing content automatically.