Deepfake Video Call Scams

Deepfake video call scams use real-time AI face and voice synthesis to impersonate a known, trusted person during a live video call. The technology overlays a fabricated facial image and cloned voice onto the scammer's actual camera feed, so that the recipient of the call sees and hears someone who looks and sounds like an executive, colleague, family member, or other trusted figure.

This represents a critical escalation in fraud capability because video calls were previously relied upon as a strong verification method. The cultural assumption that 'I can see and hear them — it must be them' is deeply embedded in both personal and professional behaviour. Deepfake video call fraud directly exploits that assumption.

In corporate variants, finance staff, accountants, or executives are directed by an apparently present and speaking CEO, CFO, or external auditor to execute an urgent, confidential bank transfer. In personal variants, a family member appears on a video call in distress requesting emergency funds. In both cases, the visual 'proof' of identity removes the hesitation that a voice call alone might not overcome.

Preparation involves gathering sufficient source material on the impersonation target: video and audio from public appearances, interviews, social media, or internal company content such as recorded meetings or presentations. This material trains a real-time synthesis model that can animate the target's face and produce new speech from text typed or spoken by the scammer.

The call is placed through a standard video-conferencing platform or messaging app. The scammer activates the deepfake layer through a virtual camera application that intercepts the video feed before it reaches the conferencing software. The recipient's screen shows the deepfake face in a standard video window with no visible indication that it is synthetic.

The call script is crafted for maximum compliance: it establishes urgency, invokes authority, and stresses confidentiality. In business settings, the 'executive' explains that a deal, acquisition, or regulatory payment must be completed within hours and must not be discussed with other staff until it is done — both conditions serve to bypass existing approval processes. The scammer maintains the impersonation long enough to obtain a commitment to pay, then ends the call.

A follow-up email from a spoofed address reinforces the instruction and provides payment details. The combination of apparent video presence and written confirmation is designed to satisfy whatever verification impulse a target might have.

Video has been treated as a gold standard for identity verification precisely because it combines visual recognition, voice recognition, and real-time interaction in a way that seemed impossible to fake at scale. This mental model has not yet updated across most organisations and families, even as the technology to fake it has become commercially available.

The authority gradient in workplace settings is a powerful amplifier. An instruction from an apparent CEO or CFO carries enormous social weight, and disagreeing with or questioning such an instruction — even to perform a routine verification step — feels like career risk. Scammers design the call to exploit this power dynamic directly, framing verification requests as a sign of distrust or as a threat to the deal's confidentiality.

Time pressure prevents the kind of reflection that would ordinarily trigger scepticism. The combination of urgency, apparent video proof, and authority from a recognised figure creates conditions in which almost any instruction can be executed without the normal layers of scrutiny.

A finance team member receives a video call invitation from what appears to be a senior colleague and an external adviser. Both participants appear on camera and voice. The senior figure explains that a confidential acquisition is in its final hours and requires an urgent transfer to a holding account, stressing that the transaction must not go through standard approval channels due to regulatory sensitivity. The finance team member executes the transfer. The genuine senior colleague, when later reached, has no knowledge of any such call or instruction.

(On video, 'CEO') I need you to process this transfer to [account] now — it's confidential, don't discuss it with anyone.

(On video call) This acquisition must close by end of day. Process [amount] to the holding account — the legal team will follow up.

('Family member' on video) I'm in trouble abroad. I need [amount] sent urgently. I can't explain on the phone — just please do it.

(In follow-up email) As discussed on our call, please process the transfer to [account details]. Do not use internal channels for this.

(On video) Our external compliance team has flagged an issue. You'll need to move [amount] to a safe account immediately.

(On video) Don't tell [colleague] — this is part of a surprise acquisition and the deal leaks if it goes through normal approval.

Treat video presence as no longer sufficient to authorise financial transactions. Any payment instruction received over video — regardless of who appears to be giving it — must be verified through a separate, independently established channel before it is executed.

For business settings, establish a written policy: any payment request over a specified threshold requires a callback to the requester using a number from your internal directory, not a number provided during the call. No legitimate executive will object to a thirty-second verification callback. Any resistance to standard verification is itself a major red flag.

During a suspicious call, attempt to introduce a physical or spontaneous test: ask the caller to perform an unexpected action, reference a recent internal event or shared private detail that a scammer would not know, or suggest switching to a different platform. Deepfake systems have latency and can struggle with unexpected prompts.

For family calls, the same safe-word protocol used for voice scams applies: agree a private word in advance that a family member will use to confirm they are genuinely in difficulty. The safe word should be asked for casually mid-conversation to avoid alerting a potential scammer.