Blog · October 6, 2026

Labelling AI-generated ads: what to disclose

If your video ad has a generated person, place or voice in it, label it as AI-generated. It's increasingly what platforms ask for, it's what viewers expect, and it costs you very little. This guide covers why, what the main platforms and rules say in broad terms, and the things a label doesn't fix.

This is general guidance, not legal advice. Rules differ by country and platform and change often, so check the current policies where you advertise, and get advice for regulated products such as health, finance, alcohol or anything aimed at children.

Why label

• Trust. People feel misled when they find out later that the person recommending a product doesn't exist. Being upfront avoids that.
• Platform rules. The major platforms ask for realistic AI-generated content to be disclosed, and some label it automatically when they detect it.
• The law on endorsements. Consumer protection rules in many places treat a fake person giving a recommendation as deceptive.
• It's easy. A toggle when you upload, or a short line in the caption.

What the platforms ask

• TikTok Ads Manager: its ad disclaimer guidance lists AI-generated, synthetic or significantly manipulated media among the cases requiring a disclaimer. Check the synthetic-media policy and the controls available for your ad format.
• Meta (Facebook and Instagram): Meta is expanding AI info labels in About this ad to ads made with third-party AI tools when detected through industry signals. Availability and requirements can vary by region and ad category; check the current policies rather than assuming an automatic label completes your disclosure.
• YouTube: creators must disclose realistic altered or synthetic content through the upload setting, including realistic scenes that did not occur. Paid campaigns also need to meet Google Ads policies; the YouTube upload setting is one part of that review.

These policies describe different settings for organic posts and paid ads. Check the current version for the placement and country you will use before launching.

Fake endorsements and testimonials

In the US, the FTC's Consumer Reviews and Testimonials Rule guidance distinguishes fictional dramatizations and AI avatars from fake or false testimonials. There is no blanket ban on AI avatars in marketing, but an avatar does not make a fabricated customer experience acceptable, and its use can still be deceptive.

Keep a generated presenter to substantiated product facts and clearly fictional demonstrations. Do not present a generated person as a real customer describing experience they never had. A generated result is not evidence that your product works. Review the applicable advertising and transparency rules in every country you target, especially for regulated claims.

What a label doesn't fix

• A false claim. "AI-generated" on a misleading ad is still a misleading ad.
• A fake testimonial. Labelling it doesn't make it acceptable.
• A look-alike. An AI character that resembles a real celebrity or creator can mislead viewers and infringe their rights, label or not.
• Someone's likeness used without consent. Photos or a voice of a real person need that person's agreement, including for use in ads.

Voices

Use a stock voice, your own, or the voice of someone who has agreed to it. Don't imitate a recognisable voice, even with a label. If the voice is generated or cloned, include it in your disclosure.

How to label

• Use the platform's own setting when you upload or create the ad, where there is one.
• Add a short line to the post's caption, such as "AI-generated video".
• Consider a small on-screen label for ads that will be reposted or shared outside the platform, where the platform label won't follow them.

Or make the disclosure part of the ad itself, as this one does: "Guys, I just tried out fairytale video. I am literally AI."

A checklist before launch

• The ad is labelled as AI-generated, using the platform's setting where there is one.
• The person in it is fictional, or a real person who agreed in writing.
• The voice is one you have permission to use.
• Nobody in the ad claims to have used the product or describes results.
• Every claim matches your product facts and can be backed up.
• Nothing resembles a real celebrity, creator or brand other than your own.
• Extra rules for regulated products have been checked.

For hooks that a generated creator can honestly say, see 30 UGC ad hooks. For casting fictional creators, see AI actors for ads.

Staying honest in fairytale video

fairytale video is designed around these rules. The library's creators are fictional, and the studio asks you to only use a voice you have permission to use. Each video is prompted with the exact lines you approve; check the generated speech before posting. Write it for me drafts a description and lines from your product and is instructed not to invent prices, offers or claims. Review the draft against your facts. When you post, label the ad as AI-generated using each platform's own setting. More on the format in AI UGC ads.