Guide
AI video labels: TikTok, YouTube and Meta
Start with what your audience sees and hears. A generated person, an altered event and an AI-assisted script call for different decisions. Use the platform's rule, then check the disclosure on the finished post.
By LoopBatch · Checked 7 October 2026
- What you will have
- A clear disclosure decision for each destination.
- What you need
- Your final video, its audio and a record of which parts used AI.
- How long it takes
- About five minutes to review a short video.
Before you start
- Watch the final export with sound. Include generated pictures used inside an otherwise filmed video.
- Note which visuals, voices, music and words were generated or materially changed.
- Check each destination separately. A decision for YouTube does not automatically cover Meta or TikTok.
When each platform asks for a label
| Platform | When to disclose | A useful distinction |
|---|---|---|
| TikTok | AI-generated or significantly edited content showing realistic people or scenes. Read TikTok's rule. | Its guideline exempts generic text-to-speech that does not imitate a recognizable person's voice. |
| YouTube | Realistic generated scenes, meaningful alterations to real people or events, and AI music that is the video's main focus. Read YouTube's examples. | Script assistance, captions and minor aesthetic edits are treated differently from realistic synthetic media. |
| Facebook, Instagram and Threads | Meta requires its disclosure tool for organic posts with digitally created or altered photorealistic video or realistic-sounding audio. Read Meta's requirement. | A caption explaining AI use and Meta's native AI info label are separate things. |
Four examples to make the decision easier
A filmed tutorial with an AI-written outline
You filmed the demonstration and recorded the narration yourself. Using AI to organize the outline is production assistance; YouTube lists this separately from content that needs disclosure. Check whether any generated media was added later.
A realistic scene that never happened
Your Reel shows a lifelike fictional chef demonstrating a recipe. Treat the synthetic person as part of the finished video, even if the recipe is real. Use the destination's disclosure requirement rather than describing the whole clip as filmed footage.
A voice made to sound like somebody else
A generated imitation changes what a viewer believes that person said. Review both the voice and the platform rule. An AI label does not establish consent to use somebody's identity.
An obviously illustrated story
A talking paper boat is different from a convincing recording of a real event. YouTube distinguishes clearly unrealistic content from realistic changes. Do not carry that exception across to another platform without checking its own rule.
These are our examples for applying the linked rules, not platform approvals of a particular video.
Use the platform's disclosure control
In YouTube Studio, the current help page calls the upload setting AI use. Select Yes when the video meets its disclosure requirements. For TikTok and Meta, review the AI disclosure option in the publishing flow you use; app versions and account flows can differ.
A caption such as “AI-generated visual scene; recipe tested by our team” gives viewers useful context. It does not replace a required native disclosure control. Meta may also apply AI info through detected signals, and the label for some AI edits can appear in the post's menu.
How to review disclosure in LoopBatch
By default LoopBatch leaves disclosure to the creator on each post, as the platforms' own apps do: tick the disclosure when a model made or materially altered the picture, video or voice. An admin can open Settings → AI disclosure and choose I decide on each post (the default), Only when a platform requires it (LoopBatch then requires the label on a post whose media a model made or altered, which is broader than some platform exceptions, including TikTok's generic narration exception) or Every post a model touched (which also covers AI-written words over your own footage).
At approval, review AI-generated content disclosure. LoopBatch sends a supported disclosure field for TikTok, YouTube and Instagram. Its Facebook and Threads publishers currently rely on a caption disclosure; this does not guarantee a native Meta AI info label. Review the available native control when managing the post on Meta.
The steps
List the generated or altered parts of the finished video, including its voice and music.
Use the platform table and linked examples to decide which disclosure applies.
Set the native disclosure control in the upload flow, or review the supported disclosure setting in your publishing tool.
Add a specific caption explanation when it helps viewers understand what was generated.
After publishing, open the viewer-facing post and review its label, caption and audio together.
Check the result
- The disclosure describes the final media rather than only the tool used to write a script.
- The native disclosure setting and caption have been reviewed separately.
- A second destination has its own disclosure decision.
- LoopBatch's approval setting is not treated as proof that Meta displayed a native label.
If something goes wrong
- If a post that should carry a label has none, check the workspace's AI disclosure setting: under I decide on each post, the label is sent only when it is ticked on the post.
- If you see a caption disclosure but no Meta label, review the native disclosure control and the post's menu. Caption text alone does not prove the native label was applied.
- If an automatically applied label looks wrong, use the platform's own review options. Changing a filename or caption is not a reliable way to change a label.
Sources and product behavior
These sources were checked on the dates shown. Product limits and planning advice are identified separately from platform requirements.
- TikTok: Integrity and Authenticity guidelines, read 24 September 2026
- YouTube Help: disclosing use of GenAI content, read 7 October 2026
- Meta: disclosure for realistic synthetic video and audio, read 7 October 2026
- Meta: AI info labels and edited content, read 7 October 2026
Common questions
Does an AI-written script always need an AI video label?
YouTube lists script and outline assistance among examples that do not require disclosure. Review the visuals and audio separately, and check the rules of every destination. LoopBatch can optionally label every post a model touched.
Is text in the caption the same as a native AI label?
No. Caption text explains the content to viewers. A platform's disclosure control or detection process applies its native label. Use a required native control even when you also add a caption explanation.
Does a label make every AI video suitable to publish?
A disclosure tells viewers about the media. You still need to review accuracy, identity use, permissions and the destination's content rules before publishing.
Related pages
Put the plan into practice
Build a workflow you can run again.
Describe your content, review the estimate and approve the finished post before it goes out.