AI tools are now part of almost every creator’s workflow: scripts, voices, images, motion, cleanup. So a fair question comes up a lot: when do you actually have to tell people?

The short answer is that the big apps ask you to label realistic AI content, the kind that could be mistaken for real people, places or events. Using AI to help write a script or clean up your audio generally doesn’t need a label. Here’s the detail for each app, as of this writing.

This is general information, not legal advice. Rules change often; the Platform Rules reference links to each app’s official page.

The rule of thumb

Label it if a viewer could mistake it for something real.

That usually means:

  • A real person seeming to say or do something they didn’t.
  • Altered footage of a real event or place.
  • A realistic scene that never happened.
  • A realistic synthetic voice presented as a real person speaking.

That usually doesn’t mean:

  • Clearly unrealistic or animated content: cartoons, fantasy, obvious effects.
  • Minor edits: color correction, cropping, lighting fixes, beauty filters (though filters on a product’s results are a different problem; see below).
  • Behind-the-scenes help: AI ideas, scripts, titles, captions, transcription.

TikTok

  • Needs a label: AI-generated or heavily edited content showing realistic-looking people or scenes.
  • Doesn’t: small edits like color correction or cropping, and generic text-to-speech narration.
  • How: turn on the “AI-generated content” setting in the post settings before you post, or add your own clear caption, sticker or watermark.
  • Automatic labels: TikTok labels some content itself, such as videos made with its AI effects or files carrying Content Credentials. Those labels can’t be removed.
  • If you don’t: unlabeled realistic AI content may be removed, restricted or labeled by TikTok.

YouTube

  • Needs a label: a real person seeming to say or do something they didn’t, altered footage of a real event or place, or a realistic scene that never happened.
  • Doesn’t: clearly unrealistic or animated content, beauty filters, color and lighting fixes, and AI help with scripts, ideas, titles or captions. YouTube also says cloning your own voice for voiceovers or dubs doesn’t need its AI disclosure.
  • How: answer the altered or synthetic content question in YouTube Studio when you upload. You can usually change it later in the video’s details.
  • Automatic labels: YouTube may add a label itself when it detects realistic AI content.
  • If you don’t: creators who repeatedly don’t disclose can face labels added for them, removed videos, or suspension from the Partner Program.

YouTube also has a separate rule worth knowing: repetitive or mass-produced content isn’t eligible for monetization, however it was made.

Instagram and Facebook

  • Needs a label: photorealistic video or realistic-sounding audio that was digitally created or altered.
  • Doesn’t: AI that only helped behind the scenes, like script ideas.
  • How: turn on the AI label in the post’s settings before you share it.
  • Automatic labels: Meta adds an “AI info” label when it detects industry-standard signals that content was made or edited with AI.
  • If you don’t: Meta says it may apply penalties, and may add a more prominent label to content with a high risk of misleading people.

Pinterest

Pinterest shows an “AI modified” label when the creator says a Pin was made or changed with AI, or when its own systems detect it. It reads AI information in the file and uses its own detection too.

The law: the EU’s AI Act

From 2 August 2026, under Article 50 of the EU’s AI Act, if you publish a deepfake (AI-made or altered image, audio or video that looks real), you must clearly say it’s artificial:

  • Disclose it the first time people see it, in a way they can notice without special tools.
  • For clearly artistic, satirical or fictional work, a lighter disclosure that doesn’t spoil the work is enough.

Other places are adding their own rules, especially around ads and synthetic performers. If your audience is international, the safe habit is to label realistic AI content everywhere.

Ads and endorsements have extra rules

If a video is an ad or a sponsored post, the honesty rules get stricter:

  • AI “customers” or testimonials are fake reviews. In the U.S., the FTC’s rule banning fake reviews covers AI-generated ones.
  • Filters on results are false claims. A beauty filter or AI touch-up on the product being judged (skin, teeth, a before-and-after) misrepresents what the product does, labeled or not.
  • Virtual influencers are still endorsers, with the same disclosure rules as people.

A label never makes a misleading video okay

This is the most important point. A label tells viewers something is artificial. It doesn’t make it fine to:

  • Put words in a real person’s mouth.
  • Present an illustration as evidence.
  • Fake a result, a review or an event.

If a video would deceive people without the label, fix the video. The honest way to use AI in factual videos is to keep the evidence real (film the test, the product, the result) and use AI for illustration that’s clearly marked as illustration: what’s happening inside something, a place you can’t film, a reconstruction.

How to label well

  • Use the app’s setting and, if it helps, your own words too: “AI-generated illustration,” “AI voice,” “reconstruction.”
  • Label early, not only at the end.
  • Be specific where it matters: “the narrator’s voice is AI-generated” is clearer than “made with AI.”
  • Keep records of the tools you used, in case anyone asks.

Try it

Look at your last ten videos. For each, ask: could a viewer mistake any part of this for a real person, place or event that it isn’t? If yes and it isn’t labeled, add the label in the app’s settings now. Then make it part of your upload checklist.