No filming:An AI avatar or clone yourself
    Versely
    Guides

    How to label AI content on YouTube

    Tick altered/synthetic at upload. Shorts get an overlay; long-form sits below the player.

    Versely Team••7 min read

    Disclose realistic AI at upload using YouTube’s altered / synthetic (AI use) question. Photoreal Shorts then carry an overlay on the video. Photoreal long-form sits below the player, above the description. That placement is YouTube’s May 2026 change (YouTube Blog, 27 May 2026). Labels alone do not impact recommendations or monetization eligibility. This page is the upload path, the placement check, and how to contest an auto-label. It is not the inauthentic-content policy.

    YouTube still requires creators to disclose when AI has meaningfully altered or generated realistic content. Auto-detection is the backstop if you leave the question blank, not a reason to skip it.

    Tick the box at upload

    You get the first chance during the upload flow. Studio on desktop and the mobile upload both ask. The wording has been “Altered or synthetic content” and, in the May 2026 creator note, an “AI use” survey. Same job: say whether realistic AI is in the file.

    Disclose when the file:

    • Makes a real person appear to say or do something they did not.
    • Alters footage of a real event or place.
    • Generates a realistic scene that did not occur (including a photoreal talking head that was never filmed).
    • Includes AI-generated music in the cases YouTube still lists as disclosable — check the current Help Center article (YouTube Help: AI-generated content) if music is the only AI in the timeline.

    Do not disclose for:

    • Beauty filters.
    • Colour and lighting.
    • AI scripts or outlines.
    • Obvious animation or fantasy.
    • Cloning your own voice for voiceover or dubs, which YouTube has explicitly carved out.

    Procedure:

    1. Upload the finished file. Caption and crop before this if the destination is Shorts; the overlay will sit on the picture.
    2. In Details / the checks step, complete the AI use / altered-or-synthetic question. Do not skip it to “see if detection fires.”
    3. For a photoreal generate from Veo 3.1 or any other realistic row, the answer is yes.
    4. For a camera original with burned-in captions only, the answer is no. Captions are not a meaningful alteration of reality.
    5. Publish. Then open the watch page as a viewer, not as the Studio owner preview that hides chrome.

    If you use YouTube’s own GenAI tools (Veo or Dream Screen inside YouTube), the label is applied at source. You cannot take it off. That is a YouTube-internal path. A Versely export is not that path. You still answer the survey.

    C2PA in the file can also force a label. If the manifest says fully AI-generated and it survives ingest, YouTube may treat that disclosure as permanent. Do not upload a signed fully-AI file and then try to claim it is a camera original.

    Check overlay versus below-the-player

    Placement is format plus photorealism, not mood.

    Photoreal / meaningfully AI-altered (the May 2026 “main stage” label):

    • Shorts: overlay on the video. It occupies real pixels. Treat it like chrome. If you burned a caption into the same corner, the viewer sees a collision. Check safe area including this overlay, not only TikTok buttons.
    • Long-form: directly below the player, above the description. It does not cover the picture. It does cover the old habit of “the disclosure is somewhere in the description.”

    Unrealistic, animated, or only slightly altered:

    • Disclosure, if present, can sit in the expanded description. YouTube’s blog: that is not the prominent format.

    Procedure after publish:

    1. Open the Short on a phone, not only in Studio. Confirm the overlay. Confirm it does not sit on the face or the pack label.
    2. Open the long-form watch page. Confirm the label below the player. If you also wrote “made with AI” in the description, that is extra copy, not the control.
    3. If the file was meant to be animation and you got the prominent overlay, you either disclosed as photoreal by mistake or detection classed it as photoreal. Fix the disclosure in Studio if you are allowed to (next section).
    4. Do not recrop the Short to “hide” the overlay. You cannot hide YouTube’s overlay with a crop. You can only keep your own type out of its way.

    The overlay is why Shorts caption placement changed in May 2026 even if your caption style did not. Preview with the overlay in mind, then burn captions. Burning first and discovering the AI chip later is how you recaption.

    Labels will not, by themselves, limit audience or Partner Program eligibility (YouTube Blog, 27 May 2026; Help Center). The policy that actually decides whether AI video earns is inauthentic / mass-produced templated output, not this checkbox.

    Contest an auto-label — and know when you cannot

    From May 2026, if you do not specify and systems detect significant photorealistic AI use, YouTube applies a label. You get a Studio notification.

    You can update the disclosure in YouTube Studio when the auto-label is an internal-detection miss: a camera original classed as generated, a graphic classed as photoreal, a filter-only video.

    Procedure:

    1. Open YouTube Studio → the video → the disclosure / AI use control (wording follows the current Studio build).
    2. Set the status to what the file actually is.
    3. Save. Recheck the watch page in 24 hours. Do not file a public comment as the process.
    4. Keep a note of the original file and the timeline (camera vs generate). Appeals fail when you cannot say what the pixels are.

    You cannot take the label off when any of these is true (YouTube Blog, 27 May 2026; community note the same day):

    • The content was made with YouTube-internal tools (Veo, Dream Screen).
    • The file carries verified C2PA metadata indicating it is fully AI-generated.
    • Trust and Safety manually reviewed and labelled it.

    Do not strip C2PA in order to make the third-party generate “contestable.” Stripping the cooperative signal and keeping a pixel watermark is a worse position than disclosing. Disclose.

    If Studio will not let you change it, you are in one of the locked cases. Upload a true camera original as a new asset if the labelled file was the wrong file. Do not loop the same signed generate through a new title.

    Consistent non-disclosure is the path to a manual label, removal, or Partner Program trouble. The Help Center has said that for years. Auto-detect did not make hiding smarter.

    A Shorts-shaped finish

    Most AI clips in this catalog are Shorts-shaped: 9:16, under a minute, photoreal enough to get the overlay.

    Finish in this order:

    1. Generate and edit.
    2. Preview chrome including the AI overlay.
    3. Burn captions in a zone that is not the overlay and not the subscribe UI.
    4. Upload. Answer the AI use question honestly.
    5. Confirm overlay + captions on a phone.

    If the clip is a long-form chapter, the overlay is not your problem; the below-player chip is. Description text is still not the disclosure.

    A Versely export from the AI video generator does not tick YouTube’s box for you. You tick it. If you need a studio seat to finish captions before that upload, sign up. The checkbox still lives in Studio.

    FAQ

    Where does the YouTube AI label appear after May 2026?

    Photoreal / meaningfully altered: overlay on Shorts; below the player, above the description, on long-form. Animated or slight edits can stay in the expanded description. Source: YouTube Blog, 27 May 2026.

    Will disclosing AI hurt recommendations or monetization?

    YouTube says labels alone do not impact recommendations or monetization eligibility. Mass-produced inauthentic content is a different policy.

    Can I remove an auto-applied AI label?

    Often yes, in Studio, when detection was wrong. Not if the file used YouTube’s own Veo/Dream Screen, carries verified fully-AI C2PA, or was labelled after Trust and Safety review.