AI UGC ads: formats, scripts and the rules for synthetic presenters

What AI UGC ads are

UGC ads (user-generated content) borrow the look of a video a customer or creator films on a phone: a person at home, talking to camera, holding the product. AI UGC keeps the look and replaces the filmed person with a generated one, so a seller gets creator-style video without casting, shipping samples or paying per creator.

Each format makes more of the ad synthetic, and adds a risk:

Faceless UGC ads

A faceless UGC ad shows hands and the product, never a face: hands hold it up to the camera under the hook, product shots fill the middle, and the hands return for the call to action while a narrator speaks. It's hard to get right, because a model redraws the product in the hand: show the real label from your photos somewhere, and check every hand shot against the product before you publish. See our faceless UGC guide.

A 15-second structure that works

TikTok's creative guidance asks for the product's proposition in the first 3 seconds and the hook within the first 6, vertical 9:16 video at 720p or more, sound on, and on-screen text at 5–10 words per second (TikTok Ads Help, June 2025). A presenter fits that guidance if the face comes first and the product does the work in the middle:

1. Step 1

   Face with the hook, 0–3 s

   The presenter looks into the camera and says one line about the problem or the result. A face in the first frame reads as a creator, not an ad.

2. Step 2

   Product shots, 3–12 s

   Cut to the product while the voice carries on: two or three shots of 2–4 seconds each, the product in use, its size, the detail that sells it. Your own photos keep the product true.

3. Step 3

   Face with the call to action, 12–15 s

   Back to the presenter for one clear ask, then an end card with the product's name.

4. Step 4

   Captions and the label

   Burned-in captions for viewers who watch muted, and an on-screen AI disclosure. Keep both, and any price or logo, out of Meta's unsafe areas: the top 14%, the bottom 35% and 6% on each side of a Reels ad (Meta Ads Guide).

At a conversational pace, 15 seconds holds about 35–40 spoken words: a hook of under ten words, two sentences about the product, and a short call to action. A presenter who has to rush sounds synthetic sooner than one who pauses.

Spokesperson or first person: label the dramatization, keep results out

The FTC's Endorsement Guides treat as an endorsement a message consumers are likely to think reflects the opinions or experience of someone other than the advertiser, and they cover endorsers that only appear to be an individual (Federal Register). An ad that presents endorsers as actual consumers should use actual consumers, or clearly and conspicuously disclose that they aren't (16 CFR §255.2(c)). The 2024 rule on reviews and testimonials goes further and bans fake testimonials outright, naming AI-generated ones (FTC). The same rule's guidance says there's “no blanket prohibition” on AI avatars in marketing (FTC Q&A).

Put together: an AI presenter can speak for the brand, as a spokesperson or an announcer does, or in the first person as someone who uses the product, but only as a disclosed dramatization: a label that says the speaker is an AI actor and not a real customer, for the whole video, in the ad's language, opaque and readable on a phone. The dramatization can't carry what only a real customer could know: results, a claim to be a real customer, or other people's praise. The FTC's FAQ draws the line: “You can't talk about your experience with a product if you haven't tried it” (FTC).

Every line on the right is a fact your product page backs up, said as experience. If the page doesn't, leave it out: the advertiser answers for it.

Disclosure: what each platform and law asks

Our guide to labeling AI-generated ads covers each platform's label. For ads with an AI person, these rules bite:

Sources for the table: TikTok Ads policy and its built-in disclaimer, which adds a text label at the bottom of the in-feed video; Meta newsroom; 16 CFR Part 255 and the FTC's rule; New York GBL §396-b; EU AI Act, Article 50(4) and (5).

New York's law doesn't prescribe the wording, placement or duration, and none of these rules require a specific one, so pick a label viewers can't miss: a short line such as “AI actor · not a real customer” on screen for the whole video, plus the platform's own AI toggle when you upload.

What gives AI UGC away

Viewers notice. In Animoto's survey of 460 US consumers and marketers (September 2025), 82.6% said they'd watched a video they suspected was AI-generated, and 36% said AI videos lower their trust in a brand. The signs they named were robotic gestures (67%), an unnatural voice (55%) and a lack of emotional tone (51%) (Animoto). NielsenIQ's tests found AI-generated ads “less engaging” and more “annoying,” “boring,” and “confusing,” with weaker memory activation even for high-quality ones (NielsenIQ, December 2024).

Hiding the AI isn't the fix: the rules above require disclosure anyway. What you control is how much of the video depends on the synthetic face:

• Keep gestures small. A presenter who mostly talks, nods and holds still gives fewer of the robotic movements viewers spot first.
• Let the product carry the middle. Real product footage covers most of the video, so the face is on screen only for the hook and the ask.
• Match the voice to the face in age and tone, and keep the script conversational: short sentences, no lists read aloud.
• Disclose in words viewers read. A clear line of text says what they're looking at; an icon most people don't recognize doesn't.

Products to keep presenters away from

Some categories turn any person on camera into an implied testimonial: whatever the script says, viewers read the presenter as proof the product works, and these are the claims advertising rules watch most closely. For them, use a product-only video with facts from your label:

• health products, supplements and anything ingested or applied for an effect;
• weight-loss and fitness-result products;
• skincare and cosmetics sold on results (a pack shot is fine; a face that “got clearer” isn't);
• hair-loss and sexual-wellness products;
• financial, crypto and gambling offers;
• alcohol and adult products.

Making AI UGC ads with ProductToReel

ProductToReel, which we make, builds the structure above from your product link or photos. It writes three free storyboards with different hooks in about 20 seconds, without sign-up. Choose Presenter + product and one of 16 synthetic presenters says the hook and the call to action to camera while your own photos, animated, fill the middle, for 2 credits a video. The presenter can also be an AI avatar of you. With Product in hand, the presenter also holds your product: you buy stills of them holding it for 1 credit, pick one, and the video costs 2 credits, 3 in all. With Hands only, AI hands hold your product with no face while a narrator speaks for your brand: looks 1 credit, then the video 2 credits, 3 in all. Videos are 12–20-second 1080×1920 MP4s with captions, music and an end card; a presenter video is ready in 3–4 minutes.

The rules in this guide are built in: presenters and narrators speak for your brand, never as a customer and without claiming results, a script that drifts into a customer's words or a claim is rewritten, and the categories listed above get product videos instead. Every video carries C2PA Content Credentials, but platforms may not label from them, so switch on the platform's AI label when you publish. You can change a finished video in words and download it with or without captions. More on the formats on our AI UGC generator page.