Arup loses $25 million to AI deepfake impersonation of CFO
In January 2024, engineering firm Arup was targeted by a sophisticated deepfake attack. Fraudsters impersonated the CFO and colleagues via a video call to steal $25 million.
AI-generated video and audio were used to impersonate executives during a live video call to authorize a fraudulent transfer of $25 million.
Key facts
- What
- In January 2024, engineering firm Arup was targeted by a sophisticated deepfake attack.
- Incident date
- Jan 1, 2024
- Who
- Arup
- Failure mode
- Brand & Safety Incident
- AI surface
- Voice Agent
- Severity
- High
What happened
Fraudsters used AI-generated video and audio to impersonate Arup's CFO and other colleagues during a video conference call. This deception led a staff member to transfer $25 million to bank accounts in Hong Kong.
What broke inside the model
- 01 · TriggerA user prompts the model in public view.
- 02 · Model stepThe model produces unsafe or off-brand output.
- 03 · Control gapNo filter holds the line before publish.
- 04 · FailureThe output goes public unchecked.
- 05 · ConsequenceA reputational or safety incident lands.
A contained signal crosses into output that goes public.
The failure was in the identity verification process during virtual communication. Attackers used generative AI to bypass visual and auditory trust, exploiting the lack of a secondary out-of-band confirmation for the financial transfer.
What it cost
Sources
- PressArup revealed as victim of $25 million deepfake scamcnn.com
- PressScammers siphon $25M from engineering firm Arup via AI deepfakecfodive.com
- PressDeepfake AI Cybercrime Arupweforum.org
Cite this entry
https://failureindex.ai/failures/arup-loses-million-deepfake-impersonation-cfoAI Failure Index. "Arup loses $25 million to AI deepfake impersonation of CFO" (FI-0512). Realm Labs. https://failureindex.ai/failures/arup-loses-million-deepfake-impersonation-cfo (indexed Jun 16, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0512. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
Realm watches the model's internal state for the signature of unsafe or off-brand generation and can block or reroute the output before it becomes public, in real time rather than after it has been screenshotted.