What UGC ads and AI UGC ads are, and what the rules ask of them

What UGC ads and AI UGC ads are

A UGC ad is an ad made from, or made to look like, an ordinary person's own video. UGC stands for user-generated content.

The phrase covers two things. In the first, a brand takes a post that a customer or a creator really made and pays to run it. TikTok's Spark Ads work this way. In the second, the brand makes the video itself and borrows the look: one person, a phone-shaped frame, a plain room, a product in hand.

An AI UGC ad is the second kind with the person generated. The face and the voice are synthetic, and the words are a script the advertiser wrote.

Neither phrase is a legal term. We found no regulator that defines either one, and the platform help pages we read use "UGC" without defining an ad format by that name. The meaning given here is trade usage. So the rules below never mention "AI UGC". They talk about testimonials, endorsements, synthetic media and likeness, and an AI UGC ad can touch all four.

This page is a reading of public sources, not legal advice.

Is AI UGC legal in the United States

No federal rule we read bans an AI-generated person in an ad. The staff of the Federal Trade Commission say in their guidance on the reviews rule that it has no blanket prohibition on AI-generated avatars in marketing.

What the rule does prohibit, since October 2024, is a testimonial that materially misrepresents, expressly or by implication, that the person giving it exists or that they used the product. The Commission's notice says reviews written by AI are covered. An invented customer describing an experience nobody had falls inside the rule's wording.

The Commission's older Endorsement Guides point the same way. An ad that presents people, expressly or by implication, as actual consumers should use actual consumers, or disclose clearly that the people in it are not.

So the question these rules put to an ad is not whether the person is synthetic. It is whether the ad tells the viewer, in words or by how it looks, that a real customer is speaking from experience. A presenter who is plainly a spokesperson and explains what a product does stands at one end. "I have used this for a month and my skin cleared up", said by a person who does not exist, stands at the other.

Much AI UGC sits between the two, because the format is built to look like a customer's own video and the look alone can imply a customer. For that middle the rules give one test, what the ad implies to the viewer. We found no enforcement action on an AI avatar that shows how the Commission applies it.

We found no federal duty to label an ad as AI-made only because the person in it is synthetic. New York has added a state one. Its law requires an ad to disclose conspicuously that a synthetic performer appears in it, where the person behind the ad has actual knowledge of that. The bill became law in December 2025 and set its own start on the one hundred eightieth day after that, which falls in June 2026. We did not survey other states.

Is AI UGC legal in the European Union

The transparency article of the AI Act has applied since 2 August 2026. It puts duties on two parties.

The provider of a generative system must make its output detectable as artificially generated, in a form a machine can read. According to the consolidated text, an amendment gives providers of systems already on the market until 2 December 2026 for that.

The deployer, meaning whoever uses the system for anything other than personal purposes, must disclose that a deep fake has been artificially generated or manipulated, clearly and no later than the first time someone sees it.

Whether a fictional presenter counts as a deep fake is the open point. The Act's definition speaks of content that resembles existing persons. The guidelines the European Commission published in July 2026 read that to include a realistic AI-generated human avatar, and say it is enough to resemble a person who could plausibly exist.

The same guidelines say two more things that matter for an ad. The deployer's label must be one that people can perceive, and the provider's machine-readable mark does not stand in for it. And an ad may or may not count as an evidently creative or fictional work, which carries a lighter disclosure. That is decided case by case.

Guidelines are the Commission's reading and not the law itself. By their own covering note they apply only once formally adopted in every language, and we could not confirm that this has happened. We found no decision by a national authority on an ad of this kind.

For the European Union we read the AI Act and these guidelines, and nothing else. Other European rules bear on a fake customer in an ad, the rules on unfair commercial practices among them. We did not research them for this page.

What TikTok, Meta, YouTube and Google Ads ask of AI UGC

TikTok's Community Guidelines require creators to label AI-generated or significantly edited content that shows realistic-looking scenes or people. The label can be TikTok's own or a clear caption, watermark or sticker. The current guidelines took effect in September 2026. Some content is barred even with a label, a private person's likeness used without consent among it. For paid ads, TikTok's advertising policy says an ad with AI-generated content that has not been disclosed will be rejected or restricted. The guidelines also list fake reviews among deceptive behaviours. We did not read that part, so we cannot say how TikTok treats an invented customer.

Meta requires people to disclose organic posts with photorealistic video or realistic audio that was digitally created or altered. For ads, Meta labels the ones made with its own generative tools, and in June 2026 said it was starting to detect ads made with other tools through industry-standard signals. In the pages we read we found no general rule that makes an advertiser declare AI in an ordinary commercial ad. We did not read Meta's Advertising Standards or its Business Help Center.

YouTube requires creators to disclose realistic content that is AI-generated or meaningfully altered when they upload it.

Google Ads added an AI label setting in July 2026 and permits AI labels inside the ad itself. Google says that laws in the European Union, India and New York require labels on certain AI-made ad assets. In campaigns aimed at those places, the assets an advertiser designates as AI-made carry a visible overlay. Google also says that using the setting does not guarantee compliance with those laws.

These pages change often. We read all of them on the date given above the source list at the end. The date beside each source is the one the page itself carries.

A real person's face or voice in an AI UGC ad

All of the above is about people who do not exist. A real, identifiable person is a different matter. In the two states whose law we read, using that person's likeness or voice in advertising takes their consent.

California law makes a person liable for knowingly using another's name, voice, photograph or likeness in advertising without prior consent. Tennessee's law, in force since July 2024, gives every person a property right in their name, photograph, voice and likeness, and counts a simulation of the voice as the voice. Law on this differs by state, and we read only these two.

TikTok's rules bar using a real person's likeness without permission as well.

Neither law we read makes an exception for a likeness generated with software. Clipwright's acceptable use policy says what counts as consent on this service.

Does AI UGC work

We found no public study that compares AI creator-style video ads with human-made ones and says how it was done. That includes the platforms. What circulates is marketing from the companies that sell the tools.

The nearest evidence is about still images, and we read it at one remove: a university's summary of one paper and the abstract of a second, a working paper. Both report that fully AI-generated ad images drew more clicks than human-made ones in the settings studied. Each carries a warning. In the first, telling viewers the ad was made by AI lowered click-through by about a third. In the second, the advantage held only while the image did not look AI-made.

Read together with the labelling rules above, that is an uncomfortable pair of facts: some of those rules require a label, and in one study a disclosure cost clicks. We found no measurement of that for video. The way to find out for a given product is to run both versions and count.

Making AI UGC ads with Clipwright

Clipwright renders a clip from your script and returns the file. Clipwright does not post anywhere. So a label that a platform asks for is set by whoever uploads the clip, and so is a disclosure that the law asks for where the ad runs, the European and New York ones described above included.

Every finished video carries a mark that a machine can read, saying it was generated. It carries no visible label. Under the Commission's guidelines above, the machine-readable mark does not replace a label that viewers can see, and adding that label falls to whoever publishes the clip. What exactly is marked is on the AI disclosure page.

The script is yours, and Clipwright does not write it for a clip with a presenter. Under the rules above, what separates a presenter from a fake testimonial is what the ad says and implies, and that is decided in the script. A real person's photo may be uploaded only with that person's consent, as the acceptable use policy sets out.

A clip with a talking actor costs 30 credits for each second of finished video. Credits come in a pack at $10 for 1000 credits, before tax, and a new account starts with 420 trial credits. We did not research what creators or agencies charge for a UGC ad, so this page gives no market price.

This page is a reading of public sources on the date shown, not legal advice. For a campaign that matters, show the script to a lawyer who knows advertising law where the ad will run.

Sources read on 2026-10-08.