A fake endorsement advert uses a public figure's name, image or cloned voice to recruit people into an investment scam, usually through a messaging group. Japan's police record it as the single largest route into investment fraud — 3,202 cases and ¥44.09 billion through November 2025, 41% of those losses. These are paid, targeted adverts, not rumours.
A face you recognise, in a paid advert, endorsing an investment they have never heard of.
Sometimes it is a real clip recut so the words point somewhere else. Sometimes it is a fabricated news article with a real masthead. Increasingly it is a cloned voice over a synthesised face, saying something the person never said in a recording that never happened.
The advert does not take any money. Its entire job is to move you into a group chat, and everything after that is the fake platform scheme.
Almost nobody measures this. Japan does, and the number is startling: 3,202 cases and ¥44.09 billion through November 2025 from banner-style advertising alone — 41% of all the money lost to Japan’s social-media investment fraud, from one channel.
There is no reason to think the proportion is lower anywhere else. Only the counting is missing.
An economist, a broadcaster, a well-known investor. The choice is deliberate: someone whose actual job is giving financial opinions, so the endorsement is not out of character.
Real footage recut, a fabricated article with a real newspaper’s masthead, or a cloned voice. The FTC’s own framing is that AI deepfakes threaten to turbocharge impersonation fraud.
Placed through an advertising platform, targeted at people who engage with financial content, and measured. This is media buying, not a rumour.
Where it could have stopped
Search for the claim on the person’s own verified channels. A real endorsement appears where the endorser controls the message; a fake one appears only in the advert. Most of the frequently impersonated have publicly denied it — in Japan, several did so inside a National Police Agency campaign.
LINE, WhatsApp, Telegram. Japan’s police record adverts and direct messages leading to LINE investment groups as the country’s single largest route into investment fraud.
An assistant, an analyst, members posting screenshots of gains. The famous person is quoted constantly and never appears, which is explained as their being too busy.
A deposit, a rising number, and a small withdrawal that works. It is paid from other people’s money and it is the last true figure on the screen.
A withdrawal tax, a compliance fee, a verification hold. Meanwhile the advert is still running, with a different face.
Credibility is borrowed, not built. Every other investment scam has to construct a reason to be believed. This one starts with somebody the viewer already trusts, and spends nothing to get it.
The endorser is plausible. The impersonated are usually people who genuinely do give financial advice. There is nothing incongruous about them recommending an investment.
Repetition reads as corroboration. Seeing the same face across several platforms feels like independent confirmation. It is one media buy.
The old tells have expired. Reverse image search, lip-sync artefacts, a stiff voice — all were useful and are becoming unreliable. What survives is checking whether the claim exists where the person controls the message.
The platform’s presence implies vetting. An advert served by a major network feels reviewed. Automated review at that scale is imperfect, and the fact that somebody paid to reach you means only that they are organised.
And the group chat completes it. Once inside a closed group where most participants are the operation, everyone agrees, everyone is making money, and nobody can tell how many of them are real.
Two halves, and they are usually run by different people.
The advertising half buys placement, produces the creative, and is measured on cost per recruit. It uses stolen likenesses, fabricated news mastheads and cloned voices, and it rotates faces as each one becomes known.
The fraud half is the compound infrastructure documented on the pig butchering page: the group chats, the platform, the operators who manage the relationship for weeks.
What the data supports about the join between them is scale. In Japan, 3,202 cases from banner advertising in eleven months — 39.0% of SNS-type investment fraud cases and 41.2% of the losses, running alongside a separate romance-led route that produced 4,992 cases and ¥47.95 billion.
The regulatory position is unfinished. In the US the FTC’s Impersonation Rule, effective April 2024, makes it illegal to falsely pose as a business or misrepresent an endorsement, with penalties to $53,088 per violation — and US impersonation losses reached $2.95 billion in 2024. The fake reviews rule of August 2024 covers AI-generated testimonials and celebrity disclosure. But impersonating an individual is a proposal rather than a rule, and a deepfaked person saying something they never said sits in that gap.
And a supply chain exists. In Operation AI Comply the FTC charged a company whose product generated detailed reviews containing specific material details unrelated to anything the user had entered. The tooling for fabricated endorsement is itself a product.
The FTC's Impersonation Rule took effect in April 2024 and makes it illegal to falsely pose as a business or a government body, or to misrepresent an endorsement, with civil penalties of up to $53,088 per violation. In its first-year review the FTC recorded $2.95 billion in impersonation losses in 2024 and warned that AI-generated deepfakes threaten to turbocharge the problem.
Japan's National Police Agency counts SNS-type investment fraud by how the victim was first reached, and the largest single route is a banner advertisement using a celebrity's name — 3,202 cases and ¥44.09 billion through November 2025. The NPA's response was to run counter-campaigns alongside the named public figures whose likenesses had been used.
In August 2024 the FTC finalised a rule banning fake reviews and testimonials, including AI-generated ones, and requiring disclosure of material connections behind celebrity endorsements. A month later it announced Operation AI Comply, five cases against businesses selling AI-powered promises — including one whose product generated detailed reviews unrelated to anything the user had input.
Operation AI Comply, announced in September 2024, brought five FTC actions against businesses selling AI-boosted promises. Three were storefront schemes — Ascend Ecom, FBA Machine and Ecommerce Empire Builders — alleging losses over $25 million, over $15.9 million, and up to $35,000 per consumer. The Commission's summary: there is no AI exemption from the laws on the books.
When you see the advert. Do one search: the person’s name plus the platform’s name, on their own verified account and in the press. A real endorsement is on the endorser’s own channels. A fake one is only in the advert.
Then report the advert to the platform, and — if you can identify them — tell the person being impersonated. Most of the frequently impersonated maintain public denials, and several in Japan did so inside a police campaign; they generally want to know when a new one appears.
If you are already in a group. Ask for a live video call with the named expert, right now. It will not happen. Then try to withdraw everything, today, and watch what the platform does — that is the fastest test available and it cannot be answered with a story.
Do not deposit again to release a withdrawal. The tax, the fee and the verification hold are the exit, not an obstacle.
If you have paid.
Not to the advertiser. The advert is a cost, and a real one — placement is paid for through ordinary advertising platforms, with ordinary billing.
The deposit follows the route on the fake platform page: to an address controlled by the operation, converted to a stablecoin, and dispersed across many wallets. Early withdrawals are paid from later deposits, which is why they stop working exactly when the balance becomes worth keeping.
The part specific to this scheme is the economics of the funnel. An operation buying advertising has a measurable cost per recruit and a measurable return, which means it behaves like a marketing business: it tests creatives, rotates faces as each becomes known, and scales spending where the return is highest.
That is why fake endorsement adverts do not disappear when one is taken down. The face is a media asset, and there are always more of them.
The other half of this story
Our sibling site Clean on Paper explains where the deposit goes next — the movement between chains and assets that separates the money from the account that sent it.
Go to the person's own verified channels and look for the claim. A real endorsement appears on the endorser's own account, on the company's site, and in press coverage. A fake one appears only in the advert — and most of the frequently impersonated have publicly denied it, some of them alongside their national police force.
It probably was them, recut — and increasingly it is a clone. The FTC's position is that AI-generated deepfakes threaten to turbocharge impersonation fraud, and the useful test is no longer whether the face and voice are convincing. It is whether the claim exists anywhere the person controls.
Review is automated and imperfect at the scale these platforms operate. What the advert being paid for does tell you is that the operation is organised enough to buy media and target it — which makes it more dangerous, not less.
Because a closed group is outside every moderation and reporting mechanism a platform has, and because consensus can be manufactured completely inside it. Japan's police record adverts and direct messages leading to investment groups on LINE as the country's single largest route into investment fraud.
In the US the FTC's Impersonation Rule makes it illegal to falsely pose as a business or misrepresent an endorsement, with penalties up to $53,088 per violation, and the fake reviews rule of August 2024 covers AI-generated testimonials. Impersonating an individual is a less settled question — the FTC has proposed extending the rule to cover it.
It is the reverse. The face is there precisely to carry credibility the company cannot earn. If you cannot verify the company independently — regulator's register, operating history, a real address — the famous name adds nothing except the reason you kept reading.
Report it
Reporting is what produces the enforcement data on this page. Find the right agency and phone number for your country on the report page. If money moved in the last few hours, call your bank first.