Last updated 7 September 2026
AI Content & Disclosure
Everything this makes is synthetic. What that means for provenance, and where you're legally required to say so.
Everything FireplaceUGC produces is synthetic. No presenter in the output is a real person performing your script, no scene was filmed, and no testimonial in it is a real customer’s account of using your product. This page states that plainly because the alternative — an audience discovering it later — is how trust in a category gets spent.
1. What exactly is synthetic
- The presenter — generated from a licensed performer likeness. See Actor Likeness & Consent.
- The voice — synthesised, not recorded for your script.
- The performance — the presenter never said your words.
- The scene — generated, not filmed. A kitchen in the background is not a real kitchen.
- Any depicted result — if a video shows a product being used and something happening, that is a generated depiction, not evidence.
What is not synthetic is your product: we generate around the product image you supply. But a generated video is never proof that your product performs as depicted.
2. The testimonial problem
UGC-style advertising borrows the visual grammar of a real person recommending something they bought. That is the format’s power and its hazard. A synthetic presenter saying “I’ve been using this for three months and my skin cleared up” is a fabricated testimonial, and in most advertising regimes a fabricated testimonial is deceptive regardless of how it was produced.
The line we ask you to hold: a synthetic presenter may describe your product, present its features, and make claims you can substantiate. A synthetic presenter must not narrate a personal experience they did not have, report a result they did not obtain, or be presented as a real customer. Our Acceptable Use Policy prohibits the latter.
This is a genuine constraint on what you can write, and we would rather say so than let you find out from a regulator.
3. Your disclosure obligations
You are the advertiser. The disclosure duty is yours, and it varies by market and by platform:
- Advertising regulators in several jurisdictions treat undisclosed synthetic performers in testimonial-style ads as misleading.
- Platform policies on Meta, TikTok, YouTube and others increasingly require you to flag AI-generated or manipulated media when you upload it. Most provide a toggle at upload. Use it.
- Statutory regimes — the EU AI Act imposes transparency obligations on deep-fake-style synthetic content, and several US states require disclosure in specific contexts. Political advertising is separately regulated almost everywhere, which is one reason we prohibit it outright.
We cannot make these determinations for you. We can tell you that “nobody will be able to tell” is not a compliance position.
4. What we do on our side
- Provenance metadata. We are adding machine-readable content credentials that mark output as AI-generated; until that ships, exported video carries no such marker, so do not rely on one being there. Even once it does, some platforms strip metadata on re-encode, which is a reason to use the platform’s own AI disclosure toggle rather than rely on ours surviving.
- We never claim otherwise. We will not market the output as real footage, and we will not offer a feature whose purpose is to defeat AI-content detection or strip provenance data.
- Moderation. We screen generations against the Acceptable Use Policy. It is imperfect, and it runs before your review, not instead of it.
5. Known limits of the technology
Being straight about this is part of the disclosure:
- Output is probabilistic. The same request twice gives different results, and some are unusable.
- Lip-sync, hands, text-in-frame and product labels are the common failure points. Check them before you spend money on distribution.
- The model may generate a plausible-sounding claim about your product that is not true. It has no knowledge of your product beyond what you gave it. Read every script.
- Accents, names and cultural cues can carry stereotype. If output reads that way, report it — we treat that as a defect, not taste.
6. Questions and reports
To report output that appears deceptive, misuses a likeness, or carries stereotyped depiction, write to hello@fireplaceugc.com. Reports from people who are not customers are welcome and are handled the same way.
Questions about this document? Write to hello@fireplaceugc.com. If something here contradicts another page in this set, tell us — a conflict between our own policies is our problem to fix, not yours to interpret.