Threat Intelligence AI Fraud Prevention

Voice deepfakes and CEO fraud: how scammers use AI-cloned voices to divert payments

AI voice synthesis can replicate an executive's voice from short samples of public audio. Finance teams are taking calls that sound very like their CEO or CFO and authorising urgent wire transfers. Below: how the attack works, why it succeeds, and the controls that reduce the risk.

11 August 2026
Reviewed 9 November 2026
9 min read
Key takeaways
  • Voice synthesis tools can clone a voice from short audio samples, using recordings from earnings calls, interviews, or YouTube videos
  • The attack pattern combines cloned voice with social engineering: urgency, secrecy, and authority bypass normal verification instincts
  • A callback to a known corporate number, not a number provided in the call, blocks most of these attempts before money moves
  • Dual authorisation for payments above a threshold, and for any new payee or change of bank details, means no single person can move funds alone
  • If a suspicious payment has already gone, call your bank straight away on its official number: the faster you report it, the better the chance of recovering funds
  • The hardest training task is convincing staff that recognising the voice is no longer enough to trust a call

How the attack works

Business email compromise (BEC) fraud has relied on impersonation for years: a spoofed email from the "CEO" instructing the finance team to transfer funds. Voice deepfake attacks add AI-generated audio to that social engineering foundation, removing the last safeguard many finance teams trust: "I'd recognise his voice."

The technical barrier is low. Open-source voice synthesis models and commercial services can produce a convincing voice clone from short audio samples. For executives at public companies, that audio is available through earnings calls, conference presentations, and media interviews. For executives at smaller businesses, LinkedIn video posts, recorded webinars, and company promotional videos provide the same raw material.

The attack follows a predictable script. The caller contacts a finance team member, poses as the CEO or CFO, and explains that a sensitive acquisition, regulatory matter, or supplier dispute requires an urgent payment today. The recipient is asked to bypass normal approval steps given the confidential nature of the matter. Authority, urgency, and secrecy work together to suppress the verification instinct.

Cases that show the pattern

In 2019, the CEO of a UK energy subsidiary took a phone call from someone he believed was the chief executive of the firm's German parent company. The caller's voice, accent, and speech patterns were convincing. He transferred approximately £200,000 to a supplier in Hungary within the hour. The money moved through a series of accounts and was never recovered. It was one of the first publicly reported cases of AI voice cloning used in CEO fraud.

In 2024, scammers tried to impersonate WPP's CEO, Mark Read. They set up a WhatsApp account using his public photo and arranged a Microsoft Teams meeting with another senior executive. On the call they used a clone of Read's voice and YouTube footage of him, while the fake Read typed in the chat. The target was an agency leader, who was asked to set up a new business in order to extract money and personal details. The attempt failed. The case shows the move from phone-only scams to meetings that combine a cloned voice, video and messaging.

In one of the best-known cases, a finance worker at engineering firm Arup's Hong Kong office joined a video conference in which the CFO and several colleagues appeared on screen and told him to authorise multiple transfers totalling approximately US$25 million. The other participants were deepfakes. According to the Hong Kong government, the conference was pre-recorded from public video clips, so there was no real interaction. The employee contacted head office through a separate channel days later, and that is when the fraud came to light.

Why finance teams are the specific target

Finance staff work in environments where speed, deference to authority, and confidentiality are normal professional values. A CEO asking them to handle something urgently and discreetly fits the pattern of routine executive communication. Fraudsters exploit professional norms, not naivety.

Red flags that precede payment fraud

Unexpected urgency, requests to bypass normal approval steps, instructions to keep the matter confidential, and requests to use a different payment method or account than usual appear repeatedly in documented cases. Any single one is unremarkable. Together, they are the pattern.

Callers also manufacture time pressure to block verification: "The transfer needs to complete before the markets close," "I'm in a meeting and need this done in 30 minutes," "I'll explain everything afterwards." Legitimate urgent transactions can survive a two-minute callback to a known number.

What to train your staff to look for

Train finance and operational staff on one core point: voice recognition is no longer a reliable authentication method. Knowing what to look for does not stop the fraud if the only available response is "this feels odd but the voice is right."

  • Unexpected contact channel. A CEO calling from a personal mobile, WhatsApp, or unfamiliar number is a signal, not proof, but enough to trigger callback verification.
  • Request to bypass normal process. Skipping dual authorisation, using a different account, or processing a payment without a purchase order or contract are all grounds to pause and verify.
  • Secrecy instruction. "Don't tell anyone about this yet" is one of the most consistent indicators of social engineering across fraud types.
  • Time pressure. Legitimate urgent payments survive a 90-second callback. Resistance to that delay is itself a signal.
  • Subtle audio anomalies. Current voice synthesis occasionally produces micro-pauses, unnatural rhythm, or background audio inconsistencies. These are not reliable on their own but can reinforce suspicion raised by other signals.

Controls that prevent payment fraud

Detection training matters but is not enough on its own. The controls that reliably prevent fraud are procedural, not perceptual.

Callback to a known corporate number

When a payment request arrives by any channel, the finance team member should end the call and dial the requester back on a number from the company directory or official records, never a number provided during the call or in an accompanying message. It is the control that most directly defeats a cloned voice, because it does not depend on recognising the voice. Make it firm policy, not a guideline, and enforce it even when the requester pushes back on the delay.

Changes of payee and bank details

Many frauds do not ask for a new transfer: they ask for an expected payment to go to a different account. Treat any new payee or change to a supplier's bank details as a high-risk event. Confirm the change through a contact already on file, never through the channel that brought the request. Separate the duties: the person who sets up or changes a payee should not be the person who approves the payment.

Risk-based dual authorisation

No single person should authorise a payment above a defined threshold; a second approver should verify the request through a separate channel. Set the threshold from risk: your typical payment values, whether the payee is new, whether the request is urgent or unusual, and whether an overseas account is involved. Unusual payments should need dual authorisation even below the threshold.

A code word as a back-up check

Some organisations give executives a verbal code to use when making requests that deviate from standard procedure. Staff agree on it in advance and never share it digitally. An executive who does not use it when asking for an exception signals that the call may not be genuine. It is a supplementary control: code words leak, get forgotten or end up shared by message. Use it as an extra signal, never instead of the callback and dual authorisation.

Payment process documentation

Recording the authorisation chain for every payment, including informal requests, creates a paper trail that makes fraud harder to conceal. Finance teams that can show who authorised what, through which channel, catch fraudulent requests before they complete.

What to do if a suspicious payment has already gone

Act fast. Call your bank on its official number, report the fraud and ask it to stop or recover the payment; the bank can contact the receiving bank, and the sooner it does, the better the chance that funds are still there. Recovery is not guaranteed.

  • Preserve the evidence: call logs, recordings, WhatsApp messages, emails, payment confirmations and the receiving account details.
  • Report it to Report Fraud (formerly Action Fraud) in England, Wales and Northern Ireland, or to Police Scotland on 101.
  • Tell the executive whose identity was used and warn the finance team about follow-up attempts.
  • Review how the request got through your controls and fix the gap before the next payment run.

Where AI makes this harder

Voice synthesis quality is improving faster than detection technology. Error rates for detection tools vary widely with language, call quality and the synthesis model used, and the figures vendors publish usually come from controlled tests. That makes them unreliable as the only filter. Real-time deepfake detection integrated into phone infrastructure exists but has not reached wide enterprise deployment.

Procedural controls are the current best defence. Organisations that rely on technology to flag deepfakes before they reach staff are asking more than the technology can deliver. Organisations that have embedded verification procedures independent of detecting the fraud stop losses regardless of how convincing the synthetic voice becomes.

Tools that detect synthetic voice in real time

Several commercial platforms offer detection built for voice deepfakes. The examples below illustrate different technical approaches; they are not a recommendation or a complete list.

  • Pindrop Pulse. Analyses call audio for signs of synthesis and liveness. Pindrop reports high detection rates on known synthetic voice models; those figures come from the vendor, and performance against new or custom models can be lower until the tool is updated.
  • Voice biometrics (for example Nuance Gatekeeper, now part of Microsoft). Compares the caller's voice with an enrolled voiceprint rather than scanning for synthesis artefacts. It needs voices enrolled in advance and was designed for contact centres. A good-quality clone could in principle pass the match, so voice biometrics works best combined with other signals.
  • Resemble Detect. An API-based classifier that technical teams can integrate into call or audio analysis workflows without changing their telephony infrastructure.
  • Human-in-the-loop verification. No detection platform on the market is reliable enough to serve as the sole control. A callback to a pre-registered number, made through a different channel from the one that carried the request, works however good the voice clone is. It remains the strongest single control, whichever tool you deploy alongside it.

Before buying a tool, evaluate measured accuracy on audio like yours (your languages and accents, phone lines, WhatsApp), the false-positive rate, latency (the alert must arrive before the payment decision), integration with your telephony or meeting platform, and how voice data is handled. Voice recordings are personal data, and when used to identify someone they are biometric data, a special category under UK GDPR. Ask the vendor for independent test results. None of these tools replaces the process controls described above.

Why the threat is accelerating

The technical barrier to voice cloning has dropped faster than most security teams expect. A few years ago, producing a convincing clone required substantial compute resource and hours of training audio. Today, commercial services such as ElevenLabs ask for around one to two minutes of clean audio for an instant clone. Earnings calls and LinkedIn video posts give attackers enough raw material without contacting the target.

Cost is no longer a meaningful barrier. Subscription plans for these services are cheap, and free open-source tools exist. That makes mass targeting viable: an attacker can build voice clones of executives across dozens of organisations and run campaigns at scale.

Real-time voice conversion tools remove the need for pre-recorded audio. Open-source variants on HuggingFace let an attacker speak into a microphone while software converts the output to match the target's voice. The processing delay can be low enough to go unnoticed in an ordinary call. The attacker needs nothing beyond free software and a microphone.

Multilingual models have extended the threat to cross-border fraud. Current systems clone accent and intonation, not just phoneme patterns, which means an attacker no longer needs to speak the target's language to produce a convincing impersonation. International CEO fraud operations are easier to mount as a result.

Call timing is part of the attack. Scammers often pick moments of pressure, such as month-end or quarter-end close and executive travel, when time pressure on payment approvals is part of the working day. Voice deepfake attackers follow the same pattern because urgency in the environment makes verification feel like an obstacle.

UK Finance data puts the stakes in context: authorised push payment (APP) fraud losses reached £576.4 million in 2025, up 19% on the previous year, according to its Fraud Report 2026. The report does not separate out losses from voice cloning, but impersonation scams are part of that total.

For a broader look at how deepfake technology is affecting identity verification, see our guide on deepfakes and identity fraud.

Ryland Deakin
About the author
Lead Consultant, Cyvra · CISM · CompTIA Security+ · MCP

Ryland has delivered cybersecurity, compliance, and IT management programmes for regulated organisations across the UK and the Netherlands for over 20 years. View full profile

Talk to Cyvra

Does your team know what to do when the CEO calls?

We help organisations build the payment verification controls and staff training that prevent fraud before it happens.