Insurance UGC cannot invent a claimant
Insurance UGC cannot invent a claimant. No synthetic crash victims. Process explainers and recorded agents only.
Insurance UGC cannot invent a claimant. It cannot invent a crash victim, a flooded kitchen that did not flood, or a grateful parent who “got their life back” after a generated total loss. YouTube’s inauthentic-content policy, as reported by Mashable on 21 July 2026, will not monetize synthetic personas presenting as human authorities on finance (alongside healthcare, law, and politics). Insurance sits in that finance bucket when a face speaks as if they settled a claim, priced a risk, or survived an accident they did not. The production constraint is the claimant and the authority: no photoreal victims you invented, no photoreal adjusters you invented. Process explainers. Recorded agents. Hands on a real policy packet.
Carriers, MGAs, brokers, insurtechs, and comparison apps all want the peer look. The peer in this category is tightly bounded, because a testimonial about a claim is a financial claim wearing a kitchen. Ads with UGC still show about 4× click-through and about 50% lower CPC, and 86% of brands say UGC lifts performance (Kapwing 2026; Billo). That lift is not a licence to mint a claimant. 92% of consumers trust a peer over a brand message in the same stack. A peer who does not exist is a fabricated review.
Human UGC still rates 81% on authenticity against 63% for AI UGC (Kapwing 2026; archive.com / creator.co). In insurance the authenticity gap is the smaller problem. The policy gap is the larger one.
No synthetic crash victims
Do not generate a claimant. Do not generate a body in a car. Do not generate a child next to a wreck. Do not generate a “typical family” in a driveway that is not theirs, pointing at hail damage you invented. Do not generate tears. Do not generate a hospital corridor as proof of a payout.
Real claimants, if you use them at all:
- A media release separate from the claim file. Settling a claim is not a TikTok grant.
- Words they are allowed to say. “This is how I uploaded photos to the app” is a process line. “They paid my claim in a week” is a result you may not be able to run, and this page will not invent a rate card for it.
- No leverage tied to the live claim. If the cheque depended on the interview, stop.
- A withdrawal path you will honour, including ads already in flight.
The FTC’s Consumer Reviews and Testimonials Rule still applies. A generated “they were there the night of the fire” is a fabricated testimonial ad. A hired actor playing a claimant has to be disclosed as an ad, and still cannot invent a loss. Many carriers should use neither.
This is ads and social, not a claim file. Do not put a generated VIN, a generated police report, or a generated medical scene in the cut. A warped plate on a generated bumper is a fraud read and a regulator’s screenshot.
YouTube Shorts overlay photoreal AI on the video (YouTube Blog, 27 May 2026). Labels do not restore monetization in the inauthentic bucket. Labels do not create a loss. Labels do not create a licence to speak as a claimant.
New York’s synthetic-performer law (N.Y. GBL § 396-b, in force 9 June 2026) requires conspicuous disclosure of generated human performers in ads that can reach New York consumers. $1,000 first, $5,000 subsequent. Insurance ads are ads. Disclosure does not create a claim. Article 50 has applied since 2 August 2026; photoreal people that would appear authentic need a first-exposure label for EU audiences. Use the Article 50 disclosure template. Then still do not run the fake victim.
YouTube's finance bucket
Mashable’s 21 July 2026 report on YouTube’s policy walkthrough names three inauthentic-content buckets that can cost Partner Program money: generic or repetitive near-duplicates; rage-bait or distressing content without a larger story; and synthetic personas presenting as human authorities on healthcare, finance, law, and politics. AI itself stays monetizable when the work is original. Do not invent a fourth bucket. Do not invent extra sensitive topics.
A generated “senior adjuster” walking a viewer through “what your policy really covers” is the third bucket. So is a generated broker in a glass office quoting a premium they do not have authority to quote. So is a generated claimant describing a payout as if it happened.
Rage-bait without a larger story is the second bucket. Crash compilation as an insurance ad, wreck footage as a hook, “this total loss will ruin you” as a fear open — that is distress used as a click, not a process explainer. Do not generate it. Do not scrape it.
Generic near-duplicates are the first bucket: the same explainer with the state name swapped, the same slideshow of scrolling policy text, the same AI voice over stock rain. Original process stills and a recorded human are how you stay out of that pile.
A talking head in a navy blazer is still a talking head. Colour does not confer an insurance licence. If the face is giving financial information about a product, the face is a person who may give it, or the face is not in the file.
Process explainers
The honest UGC in this category is often faceless:
- The app on a real phone, dummy personal data, one job completed: photo upload, FNOL start, ID card download.
- The packet, the declarations page you are allowed to show, the deductible line that is actually on that form.
- A car that is a car, parked, no invented damage, or a home exterior you have the right to film, empty of a fake disaster.
- Hands on a clipboard that exists, not a generated “claim approved” stamp.
Record one job per file. Named in the first seconds: “this is how you start a glass claim in the app,” “this is where the roadside number sits on the card.” Completed on camera. A cursor that never submits is a trailer.
What the VO may not do: promise a payout; invent a timeline; scare a parent with a generated crash; compare premiums you did not run; diagnose an injury; speak as if the speaker is the regulator.
Fair-and-balanced advertising rules, state insurance advertising codes, and product-specific restrictions do not disappear because the file looks like a story. If the category needs a disclaimer, burn it in a size a phone can read. UGC is not a loophole. Burned-in captions are how “coverage subject to the policy” survives mute autoplay. Auto-captions will hear “liability” as “ability.” The AI caption generator is for the policy vocabulary, not for outcomes.
Captions: burn the product name, the form name, the one allowed line. Caption styles after the take. Caption timing on the disclaimer, not only on the hook.
Hands matter. Pick a human when the hands matter is the photo-of-the-bumper, the tap-through, the card-out-of-the-wallet. Do not generate the mechanism of a claim. Do not generate a weather event as proof.
Recorded agents, papered claims
If you need a face, book a licensed agent or a named employee who will be identified as such, in a room that is not pretending to be the night of the loss, reading lines compliance wrote. That file is an expert ad. It is allowed to look handheld. It is not allowed to look like a peer who “just happened” to total a car.
On camera:
- Name, firm or carrier, the licence jurisdiction as the regulator would want it.
- One process: how to start a claim, what photos to take, where the card lives, how a portal upload works.
- One disclaimer compliance requires, burned in.
Not on camera:
- A generated adjuster.
- A generated crash victim.
- A generated “I switched and saved” family in a kitchen.
- Weather disaster b-roll you do not have the rights to, generated or scraped from a news feed.
Employees: disclose employment. Do not let the edit imply they were claimants. Usage rights for an agent’s face should live in employment plus a paid-media grant if the cut will run as ads. A conference recording is not automatically a Meta campaign.
The UGC video generator can assemble a recorded agent plate, a real app screen, and captions. Do not prompt it for “a relieved mom after a wreck.” Do not prompt it for “an adjuster explaining what you’ll get.”
66% of creators offer UGC services and the creator population grew 93% year on year (Kapwing 2026; creator.co). Most of them should not be asked to play claimants. 87% of brands use UGC and 16% have a strategy. The strategy here is: process as the proof, no synthetic victims, no synthetic finance authorities, claims a specimen policy already makes.
If the cut still needs a crying claimant to make sense, the brief was a testimonial you may not be able to run. Put the app on the table.
Assemble the process cut in Versely.
FAQ
Can I generate a claimant if I disclose that they are AI?
No. Disclosure says the face is synthetic. It does not create a loss, a payout, or a YouTube monetization path for a synthetic finance authority. Keep claimants real and papered, or keep them out.
Is a licensed agent in a polo a problem?
Not if they are identified as an agent or staff and they do not fake a claimant story. It is a problem if the edit implies they were paid a claim, or if they make a coverage statement compliance did not write.