Skip to content

Authority explainer · last reviewed 21 November 2026

AI deepfake scam examples from the workplace

This page is maintained by the instructor of the Udemy course linked below. Each case cites the reporting or incident record it is drawn from, checked 21 November 2026.

Short answer

An AI deepfake scam borrows a real person's cloned voice, video, or image to make a fraudulent request feel routine. It has already cost one company $243,000 over a cloned-voice phone call and another $25.6 million over a video call with deepfake participants; a third attempt, which impersonated a CEO to target a different executive at his own company, was caught before any money moved. The fix is verifying any payment or data request through a channel the caller or the call cannot control, not trusting a familiar voice or face on its own.

What counts as a deepfake scam

A deepfake scam uses AI to generate or clone a real person's voice, video likeness, or image, then uses that borrowed identity, usually someone senior, trusted, or both, to make a request for money, access, or information look like it came from them. The technology ranges from a cloned voice on an ordinary phone call to a real-time video call built from public footage of several people at once.

Three documented cases

Ordered by what happened, not by severity. Each is drawn from contemporaneous reporting or a maintained incident record rather than from a vendor's own account.

01 · Wire fraud · checked 21 November 2026

A cloned voice on the phone, and a $243,000 wire transfer

In March 2019, the UK head of an energy firm owned by a German parent company wired €220,000 (about $243,000) to a Hungarian supplier after a phone call he believed was from his German parent-company CEO, who said the transfer was urgent and would be reimbursed. The caller used AI-generated voice technology that reproduced the real CEO's slight German accent and speech pattern closely enough that the target suspected nothing. The firm's insurer, which covered the loss, disclosed the case to the press.

The Wall Street Journal, Fraudsters Used AI to Mimic CEO's Voice in Unusual Cybercrime Case

02 · Wire fraud · checked 21 November 2026

A video call full of deepfakes, and a $25.6 million transfer

In January 2024, a finance employee at the Hong Kong office of Arup, the engineering firm behind the Sydney Opera House, received an email that looked like it came from the UK-based group CFO asking for a confidential transaction. Initially suspicious, the employee joined a video call where the CFO and several other colleagues appeared to authorise the payment; Hong Kong police said every other participant on the call was an AI-generated deepfake built from publicly available footage. The employee went on to transfer roughly HK$200 million (about $25.6 million) to five Hong Kong bank accounts before the fraud was discovered. Arup confirmed to the press that fake voices and images were used.

CNN, British engineering giant Arup revealed as $25 million deepfake scam victim

03 · Blocked attempt · checked 21 November 2026

A cloned CEO voice, and a scam that did not work

In May 2024, scammers set up a WhatsApp account using a public photo of WPP CEO Mark Read to impersonate him, then used it to arrange a Microsoft Teams call with a different senior WPP executive, the actual target of the scam. On the call the scammers played a voice clone of Read alongside YouTube footage of him while impersonating him in the chat window, asking the targeted executive to help set up a new business and hand over financial details. The targeted executive and WPP's team recognised the attempt, no money or data moved, and Read later emailed staff describing exactly how the scam worked.

Entrepreneur, Hackers Try Steal Money, Personal Information From Executives at the World's Largest Advertising Company Using a Deepfake of the CEO

What reduces the risk at work

Treat a voice or face on a call as weak evidence on its own, especially when the request involves money, access, or a sense of urgency and secrecy. Verify through a second channel the caller or the call cannot control: call a known number back, check with the person directly, or confirm with someone else before acting. Agree on this verification step before a real attempt happens, not during one.

Frequently asked questions

What is an AI deepfake scam?

An AI deepfake scam uses a generated or cloned voice, video, or image of a real person, usually someone senior or trusted, to convince a target that they are dealing with that person, then uses that trust to ask for a payment, a transfer, or sensitive information.

Can a deepfake scam actually cost real money?

Yes. The cases below cost one firm $243,000 over a phone call and another firm $25.6 million over a video call; a third attempt, which impersonated one executive's voice and image to target a colleague, was caught before any money moved.

Why do these scams work even on experienced staff?

They borrow a voice or face the target already trusts and pair it with normal business pressure, an urgent payment, a confidential new deal, so the request feels routine rather than suspicious, and the usual giveaways of a scam call or email are missing.

How do you catch one before it costs you?

Verify any request for a payment, a transfer, or sensitive data through a second channel the caller or the video call cannot control, such as calling a known number back or checking with the person directly, before acting on what was said or shown.

Sources

General information, not legal advice. Last reviewed 21 November 2026.

Teach staff to question a familiar voice or face.

A 45-minute workplace baseline on Udemy covers how AI tools can be misused against you and what to verify before you act.

View the course on Udemy