A deepfaked Martin Lewis Facebook ad led to a fake adviser, a Revolut account, £5,000 of savings and £70,000 in loans. How it ran and what exposed it.
Social Engineering Examples·10 sources
In August 2023 a self-employed kitchen fitter from the Brighton area saw a Facebook video. In it, Martin Lewis, the UK's best-known consumer finance campaigner, and a well-known technology entrepreneur appeared to be recommending a bitcoin investment. Both were deepfakes. He left his details. Ten minutes later a man calling himself "Carl" rang, said he was a financial adviser, and helped him open a Revolut account. First £1,000. Then his £5,000 life savings. Then, when every withdrawal needed a bigger deposit first, four loans from four lenders totalling £70,000. Total: around £76,000. The scam ended when his son heard rap music in the background of one of Carl's calls and said: that's not a financial adviser. Status: Reported to police; "Carl" unidentified; two loans cancelled; debt outstanding.
| Step | What happened |
|---|---|
| The ad | A Facebook video using AI-manipulated footage of Martin Lewis and a technology entrepreneur endorsing a bitcoin scheme |
| The call | About ten minutes after he submits his details, "Carl" phones, posing as a financial adviser |
| The account | Carl walks him through opening a Revolut account |
| The taster | £1,000 in. Something that looks like the £1,000 coming back appears in the app. Per the Mirror's later account it was about £990, so not even a profit |
| The savings | Convinced, he transfers his full £5,000 |
| The squeeze | Withdrawals need a bigger deposit first. Four loans, four lenders, £70,000 |
| The end | His son hears rap music behind Carl's voice on a call and tells him to go to the police |
Strip out the deepfake and you've got an ordinary investment scam. What the AI changed was the bait. Instead of a stock photo and a made-up testimonial, the ad used a face the victim already trusted, mouthing words it never said. From the phone call onwards it's the pig butchering script, beat for beat. A small win. A big deposit. Then pay-to-withdraw until there's nothing left to pay with.
Tip: "you need to invest more to release your funds" has exactly one meaning. The funds don't exist.
| Item | Amount |
|---|---|
| Initial "investment" | £1,000 |
| Life savings | £5,000 |
| Four loans from four lenders | £70,000 |
| Total exposure | About £76,000 (some headlines rounded to £75,000) |
| Loans later cancelled by lenders | Two of the four |
| Debt still outstanding (later reporting) | £27,000 |
Early on, he said he was taking it to the Financial Ombudsman Service. The Independent's April 2025 follow-up describes continuing debt and doesn't establish any recovered sum. Revolut, for its part, pointed to transaction monitoring that screens "over half a billion transactions a month". Nothing in the record suggests it intercepted this one.
The kitchen fitter isn't named here. He has spoken publicly, to BBC News and on BBC Radio 5 Live with Lewis, who called him "brave and admirable" for doing so. His case fed into ITV's "Celebrity Scams: Are You At Risk?" in March 2025, which put aggregate losses from fake content featuring Lewis at around £20 million.
Not the video. Not the app. A teenager's ear for what a financial adviser's office sounds like. That's worth sitting with. The deepfake was good enough to start the chain and the fake returns were good enough to keep it going. What broke it was a mismatch between the persona and the environment, noticed by someone who wasn't emotionally invested.
The other tells were there all along: an unsolicited call minutes after clicking an ad, an "adviser" who sets up your account for you, and the pay-to-withdraw structure.
See also deepfake examples, pig butchering and the Bengaluru finance-minister deepfake.
It started with a Facebook ad showing AI-faked footage of Martin Lewis and a tech entrepreneur. A fake adviser called "Carl" then rang, had the victim open a Revolut account, took £1,000 and then £5,000, and pushed him into borrowing £70,000 to "release" returns that never existed.
About £76,000 in total, of which £70,000 was borrowed. Two loans were cancelled; later reporting put the remaining debt at £27,000.
The victim's son heard rap music in the background of a call from "Carl" and decided that didn't fit a financial adviser. The family went to the police.
No. He has said publicly that he doesn't do adverts, so any ad featuring him is a scam.
Social Engineering Examples. “Martin Lewis deepfake ad costs a Brighton kitchen fitter £76,000 (2023)”. Accessed 18 November 2026. https://socialengineeringexamples.com/deepfake-celebrity-investment-scam-revolut-2023
The scam operation likely built its fake endorsement clip from existing public broadcast and interview footage of a consumer-finance advocate and a technology entrepreneur, both frequently filmed public figures, using AI face-swap and voice-synthesis tools to produce a convincing fabricated interview-style video, consistent with contemporaneous reporting on the wave of fake celebrity investment ads.
There is no practical way for a consumer or platform to intercept synthetic-media production happening entirely inside a fraud operation before an ad is ever published; the realistic control sits downstream, at ad-platform review (Stage 2) and at consumer skepticism toward any celebrity-endorsed investment video regardless of how convincing it looks.
The fabricated video was placed as a Facebook video advertisement, exploiting the ad platform's review and targeting systems to reach a mass, unvetted consumer audience with an unauthorized celebrity-endorsement claim, a gap Meta later addressed with facial-recognition ad screening and that the consumer-finance advocate has separately sued Facebook over.
Ad-platform proactive scanning for unauthorized celebrity-likeness use in financial-services advertising, such as the facial-recognition screening Meta introduced after this wave of cases, plus faster takedown channels for public figures like the consumer-finance advocate to flag fraudulent ads before they reach mass distribution.
the kitchen fitter submitted his contact details through the ad; the operation used a call-centre-style rapid-response process, phoning him within roughly ten to thirty minutes (accounts vary slightly across BBC and Mirror reporting) of submission, consistent with organized lead-buying and lead-selling fraud call-centre operations rather than a single lone operator.
Treat any call received within minutes of submitting details to an unsolicited investment ad as a warning sign in itself, and independently verify the caller and firm through official contact details found separately, never a number or link the caller supplies.
A caller using the alias "Carl" posed as a financial adviser with over 20 years of experience and built personal rapport over multiple calls, sharing invented common interests and family details, a standard trust-building step ahead of any request for money.
Consumer education that reframes how quickly and personally a stranger builds rapport as a manipulation signal rather than reassurance, and encourages a mandatory outside second opinion from a family member or independent adviser before committing money to someone met only through an unsolicited call.
Carl walked the kitchen fitter through opening a Revolut account and taking an initial, relatively small £1,000 "investment," a typical low-stakes deposit used to build confidence before requesting a much larger sum.
Financial platforms such as Revolut can apply extra friction, including delayed release or step-up verification, to newly opened accounts receiving first-time transfers described as investments, especially where the customer says they were referred by an unsolicited caller.
The scheme showed the kitchen fitter what appeared to be his money returned inside the Revolut app interface. Later, more detailed reporting found the amount shown was actually a slightly reduced ~£990, not a profit, and it was never confirmed whether the screen reflected a genuine Revolut balance or a spoofed interface controlled by the scammer, a tactic consistent with manufacturing false confidence before extracting a larger sum.
Platform-level controls that make it impossible for a third party to spoof an in-app balance or returns screen, paired with consumer guidance to treat any on-screen "proof of returns" shown or described by an unsolicited adviser as unverifiable unless confirmed independently through the bank's own official channels.
Believing the scheme legitimate, the kitchen fitter transferred his entire £5,000 in life savings into the Revolut account.
Hard transaction limits, cooling-off prompts, or real-time fraud warnings for transfers representing a large share of a customer's total balance or described as "life savings," particularly to a newly opened account or one linked to a recent unsolicited investment referral.
The scammer withheld withdrawals unless the kitchen fitter "reinvested" larger sums, telling him he had to invest £10,000 to get his £5,000 back, a classic advance-fee and escalating-investment script that converts sunk-cost psychology into further payments.
Widely repeated consumer-protection messaging, including the consumer-finance advocate's own public warnings, that a legitimate investment never requires paying more money to unlock a withdrawal, treating that specific claim as a definitive scam indicator.
Under this pressure, the kitchen fitter took out four separate loans from four different lenders totaling £70,000 to keep feeding the scheme, an example of debt-fueled fraud that pushes victims beyond their savings into consumer credit.
Lender-side affordability and purpose checks that flag rapid, multiple loan applications within a short window, especially where the stated purpose is investment, are the realistic control here; in this case it took after-the-fact review, not real-time screening, for two of the four loans to be cancelled.
The scammer moved the transferred funds out of the Revolut account under the pretense of investing them, completing the theft.
Once funds leave the kitchen fitter's own regulated account, recovery depends on cross-institution fraud-fund tracing and mule-account detection at the receiving end, largely outside the kitchen fitter's control; the more realistic controls sit upstream, at Stages 7 and 9, preventing the transfer in the first place.
the kitchen fitter's son overheard rap music playing in the background of a call with "Carl," judged it inconsistent with a professional financial adviser's call centre, and urged his father to go to the police, ending further payments and prompting the fraud report.
The single most effective control in this case was informal: a family member's outside skepticism catching a mismatched detail during a live call. Normalizing "loop in a trusted third party before sending money to anyone met through an ad" as standard advice replicates this after every earlier institutional control had already failed.
Browse by what this case has in common with others in the library.
Fraudsters faked WPP CEO Mark Read on WhatsApp and staged a Teams call with a cloned executive voice. The agency…
An attacker sent a LastPass employee WhatsApp calls and a voicemail using an AI copy of the CEO's voice. The…
KnowBe4 hired a North Korean operative using a stolen US identity and an AI-enhanced photo in 2024. Malware started on…
Fake finance "directors" in WhatsApp groups steered 18+ US investors to NanoBit, a crypto platform that never traded. The SEC…
A spoofed call asked a Wells Fargo customer to read back a 2FA code. $100,000 left her account. The bank's…
Fake QR stickers on UK council parking machines sent drivers to cloned payment sites. 123 of 373 councils were hit…
A Singapore businessman sent at least S$4.9M after a fake Zoom "government meeting" built from real footage of PM Lawrence…
Fraudsters posing as Crelan's CEO had bank staff make dozens of transfers. €69.1M gross, €10M insured, €2.14M clawed back, nobody…
A Chicago hairstylist lost $20,000 to a Phantom Hacker scam. A caller claiming to be Bank of America said a…
Censys traced a USPS 'package awaiting action' smishing kit to 682 lookalike sites that streamed card numbers as victims typed.…
An unidentified operator jailbroke Anthropic's Claude Code with a fake "authorized bug bounty and penetration test" pretext, then defeated its…
A fake Norton bill, a fake PayPal refund and a fake PNC banker led a 76-year-old to wire $390,000 for…
MHRA found 47 crates of care-home patient records in a Harrow pharmacy's yard. ICO fined £275,000, cut to £92,000 on…
In 2019 a UK energy CEO wired €220,000 after a call that sounded like his German boss. Insurer Euler Hermes…
A Metropolitan Police officer spotted customers' passports and tax-credit paperwork clearly visible through transparent bin bags left on the street.
Anthropic says a UK cybercriminal used Claude to build ransomware they couldn't write alone, then sold kits for $400 to…
Two Scottish small businesses lost £31,000 and over £5,000 after callers impersonating bank fraud-team staff talked owners into wiring money.
Fraudsters spoofed Barclays' number and hold music, then talked a UK jeweller into installing AnyDesk. £48,451 drained; £25,650 later recovered.…
The FBI's December 2024 warning on AI-enabled fraud: AI text, fake IDs, cloned voices, and deepfake video, and what the…
A fake email posing as DTC gave JE Cleantech false wire instructions for its 2026 dividend, diverting US$794,934 and delaying…
Scammers posing as a school contractor sent Cabarrus County a bank-change request. The county wired $2.5M; $776K was frozen and…
A fake 'reactivate your MFA' email with a QR code targeted a US MSP employee in 2023, routing through a…
Impersonators posing as two Philadelphia school vendors asked to switch from cheque to ACH and diverted nearly $700,000 in 2024.…
Scammers posing as Southern California Edison threatened disconnection and demanded $806,000 in 2025. Customers paid $131,464. How the scam runs.