AI-generated video: what TikTok, Instagram and YouTube ask you to disclose

The principle every platform shares

TikTok, Meta (Instagram and Facebook) and YouTube converge on one rule: if a video contains realistic scenes, people or events that were generated or meaningfully altered with AI, the viewer should be told. None of them ban AI content. All of them dislike AI content that pretends to be real footage. A stylised 3D animation or an obviously synthetic product clip is not the concern; a realistic street scene that never happened is.

The practical consequence is simple: label what you generate, do not label what you filmed. Mislabelling in either direction is a problem — an unlabelled synthetic clip can be removed or demoted, and a real clip labelled as AI misleads viewers and undermines your own footage.

TikTok

TikTok gives creators an "AI-generated content" setting on every upload and asks that it be turned on for AI-generated or significantly AI-edited content showing realistic scenes. The label appears on the video. TikTok also reads content credentials embedded in some AI-generated files and can label those automatically.

When posting through TikTok's API, the same flag exists as a field on the post. PostingClips sets it on clips generated with AI — including edited clips that contain AI-generated footage — and leaves it off for videos you uploaded and for footage from the library.

Instagram and Facebook (Meta)

Meta shows an "AI info" label on content that was flagged as AI-generated, either by the creator or by detecting industry-standard AI markers in the file. Meta asks that photorealistic video or realistic-sounding audio created or altered with AI be disclosed, and reserves the right to add the label itself.

Instagram's publishing API exposes the flag directly. Facebook's Page publishing API does not, so the honest fallback is a disclosure line in the caption. PostingClips sets the API flag on Instagram Reels and, on both Instagram and Facebook, adds a short line to the caption — only for clips with AI-generated footage.

YouTube

YouTube requires creators to disclose "meaningfully altered or synthetically generated" content that a viewer could mistake for real — a realistic person saying something they did not say, a realistic event that did not happen, a real place altered to look different. The disclosure adds an "altered or synthetic content" label in the description, and on sensitive topics in the player itself. Clearly unrealistic content, colour correction and standard editing do not need it.

Through the YouTube API this is a boolean on the upload. PostingClips sets it on clips with AI-generated footage. Videos you uploaded and footage from the library go out without it.

X, LinkedIn, Threads, Snapchat and Bluesky

These platforms have policies against deceptive manipulated media but, as of this writing, no dedicated AI label you can set through their posting APIs. The workable approach is a short disclosure in the caption text — one line, at the end — on clips that were generated. That is what PostingClips does on all five; the line is counted against each platform's character limit so it never causes a rejected post.

What this looks like in practice

If the labelling is done by the tool, it stops being a decision you make per post. In PostingClips every clip carries its origin — generated with AI, picked from the footage library or uploaded by you — and that fact drives the TikTok flag, the Instagram flag, the YouTube disclosure and a short line at the end of the caption on every platform ("Video generated with AI."). Footage you filmed gets none of it. You never have to remember which platform wants what.

• Clip generated with AI → TikTok flag, Instagram flag, YouTube disclosure, and the caption line on every platform.
• Your own upload or library footage → no labels, no disclosure line.
• A clip edited together from several pieces is labelled if any piece was generated with AI: the labelling follows the footage, not the account or the automation.