27 October 2026 · Deepfake-enabled payment instruction fraud
Everyone else on the call was generated
In January 2024 a finance employee at the Hong Kong office of the engineering firm Arup joined a video call with people he understood to be his chief financial officer and several colleagues. Over the course of the day he executed fifteen transfers totalling roughly HK$200m, about US$25.6m.
Every other participant on the call was an AI-generated impersonation.
He had been suspicious of the original email. That is the part worth sitting with. The control he used to test it — speaking to the people involved, seeing their faces — was the control the firm had adopted precisely because email could not be trusted.
The pattern
This is deepfake-enabled payment instruction fraud: synthetic video and voice defeat the callback and verification steps firms adopted to stop email-based impersonation.
Your corporate customer asks what control now defeats this. What do you tell them?
One tap. The answer appears straight away.
