YouTube simplified its AI claims process in August 2026. If a video of yours gets flagged for depicting someone's AI-generated likeness — or you are on the other side and want to understand what the uploader sees — the response now runs through a renamed "Claims" menu and a guided choice among four dispute categories, instead of a freeform appeal. This walks the response, step by step.
One thing to fix in your head before you start: a valid AI-likeness claim is not a copyright strike. YouTube has said it carries no strike or Community Guidelines violation on its own; the consequence is that the video can be restricted from public view while the claim is resolved. So this is a distribution problem to clear cleanly, not a channel emergency — work it calmly, pick the accurate category, and produce your evidence. For the full policy background behind the flow, see the guide on [YouTube's AI claims process](/guides/youtube-ai-claims-process).
The steps
Open the video and find the Claims menu. In YouTube Studio, open the flagged video's details page. What used to be the "copyright" menu is now labeled "Claims" — YouTube consolidated copyright and AI-likeness matters there in the August 2026 update. Open it to see the AI-likeness claim, including what is being claimed and which portion of the video it applies to. Read the claim in full before you touch a response option; the right category depends on exactly what is alleged.
Confirm what the claim actually alleges. Separate the three things a likeness claim can assert: that a real person's face appears, that it was AI-generated or altered, and that it was used without permission. Your dispute lane depends on which of those is wrong (or, in the consent lane, on your having permission). Do not reach for a response until you can state plainly why the claim is disputable — the flagged material isn't AI, the person isn't in it, the use is transformative, or you have consent.
Pick the right one of the four dispute categories. YouTube offers four labeled responses. "Explicit Consent" — you have verified permission to use the person's voice or visual likeness. "Parody, Satire, or Public Interest" — the depiction is transformative, for comedy, critique, or news. "Content Not Altered or Made With AI" — the material isn't AI-generated at all (the false-positive lane). "Claimed Content Doesn't Appear" — the claimed person simply isn't in the video. Choose the one that is truthfully accurate; picking a lane you can't support wastes the review and can look like bad faith.
Gather your evidence before you submit. The consent lane is the one that needs a paper trail: a signed release, a license, or written permission that verifies the person agreed to have their voice or likeness used. For the "not AI" lane, keep the original footage, project files, or camera metadata that show the clip is authentic. For "doesn't appear," note the timestamps that disprove the match. Assemble this first so your response is complete rather than a promise to follow up.
Submit the dispute and note what happens to the video. File the response from the Claims menu with the category and any supporting detail YouTube requests. While the claim is under review the video may stay restricted from public viewing — that is expected, and it is not a strike. Resist the urge to delete and re-upload; that abandons the dispute and can re-trigger the same scan. Let the filed response run its course.
Track the outcome and escalate only if it is wrong. Watch the Claims menu for the resolution. If the dispute is upheld, the restriction lifts and the video returns to public. If it is rejected and you genuinely believe the decision is wrong — a real false positive, or consent you can now document — follow YouTube's escalation path rather than reflexively re-uploading. Keep records of the claim and your evidence in case a pattern of bad claims needs reporting.
Prevent the next claim. The durable fix is upstream. Only publish synthetic video built from a face and voice you own or have documented permission to use, disclose AI-altered content with YouTube's built-in label, and keep consent records with each project so the Explicit Consent lane is a lookup, not a scramble. A claim you never trigger costs nothing; the whole point is to keep your uploads in the "content I control" lane by default.
Common gotchas
A likeness claim is not a copyright strike. YouTube has said a valid one carries no strike or Community Guidelines violation on its own — the video is restricted while it resolves, but your channel standing is not hit. Don't panic-delete over it.
Picking a dispute category you can't support backfires. If you choose Explicit Consent with no release to show, or claim "not AI" for a video that is, the response fails and reads as bad faith. Match the lane to the truth.
Deleting and re-uploading abandons the dispute. It drops your filed response and can re-trigger the likeness scan on the new upload — you end up back where you started, minus the video's watch history.
Informal permission is not consent evidence. "They said it was fine" won't clear the Explicit Consent lane; you need a verifiable record — a release, license, or written agreement — captured before you publish.
Likeness detection is opt-in and currently visual. It only scans for enrolled creators' faces and does not yet match voices, so a claim comes from a person who signed up — not an automatic sweep of every deepfake on the platform.
Legal note
AI-likeness rights are a live and jurisdiction-specific area of law (publicity, privacy, and emerging deepfake and "digital replica" statutes). YouTube's claims process is a platform mechanism, not a legal ruling — clearing or losing a claim doesn't settle underlying rights. If you depict a real person, obtain and retain documented consent, and consult a qualified attorney for anything involving a public figure, licensed talent, or a commercial use.
Where Kompozy fits
The step you can actually systematize is the last one — prevention — and it is where a content engine earns its keep. The Explicit Consent lane clears fastest when the person in the video is one you own outright, and that is the default in [Kompozy](/). Every avatar format — [Persona Shorts](/glossary/persona-shorts), Persona HeyGen, [Persona Frames](/glossary/persona-frames) — renders from your AI Influencer persona pool: a face and voice you set up and control, never a scraped public figure. So the synthetic video you ship originates in YouTube's "content I control" lane by construction, and the consent question answers itself before a claim can be filed.
The second thing Kompozy changes is what a restriction costs you. Because a valid likeness claim only restricts one video rather than striking your channel, the real damage is the downtime — and that shrinks when re-production is cheap. From one brief, Kompozy regenerates the piece and re-fans it across the eight social platforms plus blog and email, all held to one [Persona Brief](/glossary/hyperframes) so the identity stays yours and consistent, behind a per-post review gate where you approve — and attach YouTube's altered-content disclosure label to — every upload before it goes live. A claim becomes a footnote instead of a lost week, because your queue never emptied waiting on it.
What Kompozy will not do is manufacture consent for someone else's likeness or make a bad-faith dispute stick — the discipline of owning what you depict is yours. It removes the production and governance overhead of doing that at scale. Creator ($49/mo for 2,500 credits) fits a solo creator running one owned persona across platforms; Pro ($299/mo for 18,000 credits) suits an agency or brand governing several personas and high upload volume at once; Enterprise is custom.
Frequently asked questions
What are the four YouTube AI likeness dispute categories?
Explicit Consent (you have verified permission to use the person's voice or visual likeness), Parody/Satire/Public Interest (a transformative depiction for comedy, critique, or news), Content Not Altered or Made With AI (the material isn't AI-generated), and Claimed Content Doesn't Appear (the person isn't in the video). You choose the one that accurately fits from the video's Claims menu.
Where do I respond to an AI likeness claim on YouTube?
In YouTube Studio, open the flagged video's details page and use the "Claims" menu — the section formerly labeled "copyright," which YouTube renamed in August 2026 to hold both copyright and AI-likeness matters. The claim and the four response categories are managed there.
Will disputing a claim get my video back?
If the dispute is upheld, the restriction lifts and the video returns to public viewing. While it's under review the video can stay restricted, but that is not a strike. If the dispute is rejected and you believe it's wrong, follow YouTube's escalation path rather than deleting and re-uploading, which abandons the dispute.
How do I avoid AI likeness claims in the first place?
Generate synthetic video only from a face and voice you own or have documented permission to use, disclose AI-altered content with YouTube's label, and keep a consent record with each project. That keeps your uploads in the "content I control" lane, so the Explicit Consent response is a quick lookup rather than a scramble.