Impersonation
Deepfake video call fraud: what the Arup case teaches
The best-documented AI fraud so far involved no hacked system at all. An employee was persuaded by familiar faces and voices that were not real.
Last reviewed against NCSC (UK), Anthropic and UK Government (DSIT and Home Office). Sources
What happened
The Arup fraud in outline
Building reported on 17 May 2024 that Arup had been named as the victim of a deepfake scam [8]. An employee in Hong Kong was tricked into transferring HK$200m, about £20m, after a video call that used synthetic voices and images to impersonate senior officers who asked for transfers to designated bank accounts [8].
Why it worked
The pressure points
The fraud used the ordinary structure of a business: a senior request, a call with several colleagues, a payment instruction.
Trust in what is seen
Seeing and hearing familiar people is how staff normally confirm identity. Arup's global chief information officer said attacks such as invoice fraud, phishing and deepfakes had risen and the number and sophistication of them had been rising sharply [8].
No system to break
Arup said its systems were not breached [8]. With no technical breach to detect, the control has to sit in the payment process.
Controls
What to put in place
-
1
Name the payments and requests that need a second check, such as new payees, changed bank details and urgent transfers. -
2
Define the check: call back on a number already held, or use a separate approval system. -
3
Require two approvers for transfers above a limit you set. -
4
Tell staff that a request that bypasses these steps is a reason to stop. -
5
Test the process with an internal exercise and fix what fails.
Limits of this case
It is one reported fraud, and Arup declined to share details while the investigation ran [8]. Do not read a pattern into a single case. Read it as a demonstration of what is possible.
See also AI phishing and impersonation.
Keep reading
Read next
AI intrusion
What Anthropic reported about an AI-run espionage campaign, what the NCSC expects to 2027, and what UK organisations should take from both.
Read the guide
Prompt injection
Why the NCSC says prompt injection may never be fully mitigated, and the four design measures UK teams can use to limit its impact.
Read the guide
AI tools at work
How UK organisations can govern the AI tools staff use, using NCSC guidance, the government's Code of Practice and the OWASP list of LLM risks.
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?
Many of the risks here sit in how an organisation uses AI: what tools staff run, what they can reach and who owns the risk. AI governance work sets that out so controls match the tools in use.