Impersonation
Deepfake fraud and the Arup video call
The best documented AI fraud case so far is a video call in which every other participant was fake. The defence is a process that does not depend on anyone's eyes and ears.
Last reviewed against NCSC (UK), NCSC New Zealand and UK Government (DSIT and Home Office). Sources
What happened
The Arup case
Arup said it had notified police in Hong Kong about an incident of fraud early in 2024, and Building reported the company's confirmation on 17 May 2024 [7]. Hong Kong police said an employee had joined video calls with impersonators of senior company officers, who asked for money to be sent to specified accounts [7]. HK$200m (about £20m) was transferred [7].
Why it worked
What made staff believe it
Three details matter for any organisation.
Doubt was overcome by seeing faces
According to the report, the employee initially had doubts but they were put aside because others on the call looked and sounded like colleagues [7]. Visual and audio familiarity is exactly what AI can now imitate.
The request used normal channels
The fraud arrived through a meeting, which is a routine business channel. Nothing in the setting looked unusual.
Attacks are rising
Arup's global chief information officer said it faces regular attacks, including invoice fraud, phishing and deepfakes, and that their number and sophistication had been rising sharply [7].
Your controls
Put a process around the person
Build the check into the workflow so no individual has to judge a face:
-
1
Treat any payment, bank detail change or access request made on a call as unverified. -
2
Confirm it through a second channel you already hold, such as a known number, not one supplied in the request. -
3
Require two approvers above a threshold you set. -
4
Give staff a clear, blame-free way to pause and escalate.
Remember
If the request would be wrong coming from an email, it is wrong coming from a face on a screen. A convincing call proves nothing about identity.
See also AI phishing and the exposure checklist.
Keep reading
Read next
AI phishing
What the NCSC and the 2025/2026 UK survey say about AI-assisted phishing and social engineering, and how UK organisations should adjust their defences.
Read the guide
Faster exploitation
The Five Eyes agencies and the NCSC say AI is shrinking the time between a flaw's discovery and its exploitation. What that means for patching.
Read the guide
Prompt injection
What prompt injection is, why the NCSC says it may never be fully mitigated, and the design steps UK teams can take to limit its impact.
Read the guideHow CyPro can help
Need a hand with any of this?
Patching is one job. Keeping a vulnerable system safe until it is patched, and proving it, is several more. CyPro's consultants, testers and SOC analysts can take any of them off your plate.
Vulnerability scanning
Find every internet-facing system you run, including the forgotten ones, and see which are still on a vulnerable build.
See the scanning service
Hardening and segmentation
Our consultants put the controls in place that limit the damage when something is missed: firewall rules, network isolation and tighter admin access.
More on cypro.co.uk
24/7 security monitoring
Our SOC analysts watch your systems day and night for the signs of an attack in progress: unexpected processes, new outbound connections, logins that do not fit.
More on cypro.co.uk
Penetration testing
A focused test of your systems and the controls around them, so you know they hold before an attacker finds out for you.
More on cypro.co.uk
Incident response
If you may already be compromised, our responders contain it, preserve the evidence, find out what was touched and get you running again safely.
More on cypro.co.uk
Fractional CISO for the board
A senior security lead who advises the board on what to fix first, what to accept and what to tell customers, and owns the decision record.
More on cypro.co.uk
Know your exposure
Not sure you are protected?
AI makes it quicker to find and use weak points, so it helps to find them first. A penetration test shows which of your internet-facing systems, staff processes and AI tools an attacker could actually use.