Skip to main content
Incident intelligence/SS-IR-037CASE FILE OPEN
Symbolic editorial illustration for SS-IR-037SERVANTSTACK // INCIDENT INTELLIGENCEFORENSIC IMAGE // VERIFIED FRAME
SS-IR-037 // INCIDENT REPORTReported

Arup Deepfake Fraud

AI-Generated CFO Steals $25.6 Million on Video Call

EXECUTIVE BRIEF

An employee at Arup, a multinational engineering firm, received an email requesting a confidential financial transaction.

DATA & CYBER SECURITYFINANCE & MARKETS
FAILURE CHAINTRACE COMPLETE
  1. 01TRIGGERAn employee at Arup, a multinational engineering firm, received an email requesting a confidential financial…
  2. 02MACHINE ACTIONFraud enabler
  3. 03MISSING GATEIdentity verification and dual control
  4. 04IMPACTFinancial harm
01 // INCIDENT SUMMARY

The short version

An employee at Arup, a multinational engineering firm, received an email requesting a confidential financial transaction.

02 // KEY FACTS

Case telemetry

INCIDENT
SS-IR-037
DATE
January 2024
SYSTEM
Arup Deepfake Fraud
LOCATION / SCOPE
Hong Kong
EVIDENCE
Reported
AI ROLE
Fraud enabler
HARM
Financial harm
SOURCES
1 cited record
03ENTRY POINT // WHAT HAPPENED

The event

An employee at Arup, a multinational engineering firm, received an email requesting a confidential financial transaction. Skeptical, the employee joined a video call to verify - and saw the company's CFO and several colleagues on screen, all confirming the transfer. Every person on the call was an AI-generated deepfake. The employee was the only real human in the meeting. Convinced by the realistic video and audio, the employee authorized 15 transactions totaling HK$200 million (US$25.6 million) to five Hong Kong bank accounts controlled by the attackers.

04CAUSAL TRACE // AI'S ACTUAL ROLE

What the machine did

The attackers used publicly available video and audio of Arup executives to train AI deepfake models that replicated their appearance, voice, and mannerisms in real-time on a multi-person video call. The AI-generated personas responded to the employee's questions in real-time. The technology that was supposed to connect people became the weapon that impersonated them.

Fraud enablerAutomation was a causal participant—not a decorative label for the system around it.
05BLAST RADIUS // CONSEQUENCES

Where the failure landed

$25.6 million stolen. The fraud was only discovered when the employee later verified the transaction through internal channels. Hong Kong police arrested six people. The case became the highest-profile deepfake fraud in corporate history, demonstrating that AI can now defeat the most basic human verification - "I saw them on video."

06 // EVIDENCE STATUS

Reported

Documented in the cited public record. Follow the sources for the precise evidentiary posture.

SOURCE RECORD UPDATED 2026-07-09

07 // SOURCE LEDGER

1 cited record

  1. 01
08CONTROL FAILURE // MISSING GOVERNANCE

Identity verification and dual control

The failure pattern in this case: Unverified identity or synthetic media.

09INTERVENTION POINT // HUMAN IN THE MIDDLE

The moment the path could change

A named reviewer verifies identity through a separate trusted channel before money, access, or public claims can move.

AI PROPOSES→HUMAN OWNS THE DECISION→SYSTEM EXECUTES
10CONTROL DEPLOYMENT // AUTHORITYGATE

Identity verification · dual control

AuthorityGate's framework requires out-of-band verification for high-value financial transactions - confirming through a separate, pre-established channel (phone call to a known number, in-person confirmation, hardware token). The framework treats video calls as an unverified channel for authorization. A simple callback to the CFO's known phone number would have exposed the fraud in seconds.

RELEVANT KEYSTONE CONTROLHuman-in-the-Loop ValidationHow high-risk actions route to a named subject-matter expert who owns the go or no-go decision.
12 // THE ALTERNATIVE

Autonomy is a design choice.

See the operating model that keeps AI useful while preserving human authority at consequential moments.

Compare AgenticAI and AugmentedAI →