What "AI video without avatars" means
Many of the best-known AI video tools marketed to ecommerce brands, such as Creatify and Arcads, are built around an avatar: a synthetic person, generated or cloned, who reads a script to camera. A second group of tools uses AI for the production work instead and keeps a real subject on screen. They fall into a few types:
- Catalog-driven product video. The tool reads your store's products, prices, photos and brand colors, writes scripts, and renders videos of the products themselves with captions and narration. Framewright works this way.
- Clipping and repurposing. The tool cuts your own long-form footage, such as podcasts or live streams, into short clips. See Framewright vs Opus Clip.
- Template editors with AI features. General design tools where you assemble each video yourself with AI assistance. See Framewright vs Canva.
What these share is that no one on screen is pretending to be a person who does not exist.
Why product-led video is winning on TikTok
Shoppers distrust synthetic people
The clearest data comes from Gallup. In a May 2026 survey of 3,270 US adults, 62% said it is unacceptable for businesses to use AI to create people or voices in their ads, even when the business clearly discloses it[1]. Gartner found that half of US consumers would prefer to buy from brands that do not use generative AI in consumer-facing content, and 68% often wonder whether what they see is real[2]. A second Gartner survey, published in September 2026, found 57% say AI content has made them trust brand messaging less[3].
These numbers are about AI that stands in for humans and messaging. A video that shows the actual product, at its actual price, in the brand's actual colors gives a shopper something they can verify against the store.
Platforms label synthetic media
TikTok detects and labels AI-generated content, and says it has labelled more than three billion videos as AI-generated[9]. A realistic avatar is exactly the kind of content that gets labelled. Product footage built from your own photography is not synthetic media in the same sense.
The product is what sells
TikTok's own guidance points to what drives conversions: captions with an offer, voiceover paired with the offer[10], and the value proposition in the first 3 seconds[13]. None of that requires a presenter. A product-led video puts the item, the price and the reason to buy on screen from the first frame.
What the 2026 rules actually require
This section summarises public sources; it is not legal advice. Check with a lawyer for your situation.
| Rule | In force from | Who it applies to | What it requires |
|---|---|---|---|
| New York synthetic performer law (S.8420-A / A.8887-B) | 9 June 2026 | Advertisers who know an ad uses a synthetic performer | Conspicuous disclosure. $1,000 for a first violation, $5,000 for each later one[4] |
| California AI Transparency Act (SB 942, delayed by AB 853) | 2 August 2026 | Providers of generative AI systems with over 1 million monthly users in California | Free detection tools and embedded disclosures in AI-generated images, video and audio[5] |
| EU AI Act, Article 50 | 2 August 2026 | Deployers of AI systems that create deepfakes, among others | Disclose that deepfake image, audio or video content is artificially generated[6]. Machine-readable marking for systems already on the market was given until 2 December 2026[7] |
| FTC Rule on Consumer Reviews and Testimonials (16 CFR Part 465) | 21 October 2024 | US businesses | Bans fake reviews and testimonials, which the FTC says includes AI-generated ones[8][11] |
| FTC Endorsement Guides (2023 revision) | 29 June 2023 | US advertisers and endorsers | The FTC said the update clarifies how endorsement rules apply to virtual influencers[12] |
Three practical points follow from these rules:
- A synthetic presenter in an ad is now a disclosure obligation in New York. The law covers ads where a digital asset made with generative AI appears to be a human performance by someone who is not an identifiable real person[4].
- California's law is aimed at AI companies, not brands. Its duties fall on large generative AI providers. Brands that post AI content are not the covered parties[5].
- An AI "customer" is the highest-risk use. Under the FTC rule, a testimonial that misrepresents that the reviewer exists or had the experience described is prohibited, and the FTC has said this covers AI-generated testimonials[8][11].
Product-led video that shows your real products, with AI used for scripts, layout and narration, sits outside the synthetic-performer category New York's law is built around. The EU is broader: under the AI Act, AI-generated or manipulated content that resembles real objects or places and would falsely appear authentic can also count as a deepfake[14], so AI-altered product footage is not automatically outside it. New York exempts audio-only ads[4], but AI narration inside a video is still a synthetic voice, so confirm with counsel how it is treated where you sell.
Avatar tools and product-led tools compared
| Approach | Synthetic person on screen | Disclosure exposure | Best use |
|---|---|---|---|
| AI avatar or AI actor tools | Yes | Highest: synthetic performer rules and platform AI labels | Paid-ad testing where a presenter sells the product |
| Catalog-driven product video | No | Low: real products and photography | Weekly organic posting from your catalog |
| Clipping your own footage | No, real people | Low | Brands that already film long-form content |
| Template editors | Depends on assets used | Low to medium | Teams with time to design each video |
For a tool-by-tool view, see Framewright vs Creatify and Framewright vs Arcads, both avatar-centered tools, and Framewright vs Predis, which offers avatars as one option among many.
When an avatar tool is the better choice
Avatar tools are not the wrong choice for everyone. They make sense when:
- Your category converts better with a face. Presenter-led ads are common in categories like skincare and apps, and an avatar tool can produce many variants to test. The avatar must never look like a real customer giving a testimonial.
- You are buying ads, not building an organic account. Paid creative is judged by cost per acquisition, and a fast testing loop can matter more than long-term trust signals.
- You can manage disclosure. If you sell into New York or the EU, you will need to label synthetic performers clearly and keep doing so.
Even then, never use an avatar to pose as a real customer. That is the line the FTC's fake-testimonial rule draws[8].
How to start with product-led AI video
- Start from your catalog. Your product photos, names and prices are the raw material. Choose a tool that reads them directly rather than one that needs footage.
- Keep your brand colors exact. Consistent visuals make AI-produced video look like your brand rather than a template.
- Approve every script. Keep a human in the loop for claims, tone and accuracy.
- Post on a steady cadence. Two to five posts a week is a realistic target for most small brands; our guide to the cost of TikTok content shows what that volume costs by method.
Framewright follows this approach: it reads your store's live catalog and theme colors, writes a week of scripts, and renders the ones you approve as 1080×1920 videos of your products, with no avatars.