The $35 million voice: how a cloned director fooled a UAE bank
A bank manager heard a director he had spoken to before. Emails from a lawyer backed up the call. The voice had been cloned, and $35 million was gone.
In early 2020 a bank manager in the United Arab Emirates took a call from a company director whose voice he knew. The company was about to make an acquisition, the director said, and needed $35 million transferred. A lawyer had been appointed to coordinate it, and emails from that lawyer arrived with a letter of authorisation from the director. The manager made the transfers. The director had never called.
The scheme
UAE investigators later said the fraudsters had used deep voice technology to clone the director's speech, and that at least 17 people were involved. The case became public in October 2021, when Forbes found a court filing in which the UAE asked US investigators for help tracing $400,000 of the stolen money that had passed through American accounts.
The pattern is the same one used against Fideuram five years later: a familiar voice, a professional intermediary, paperwork that agrees with the call. Each piece is ordinary. Together they read as verification.
What changes the outcome
- 01Score the voice, not the story. A detector listens to the acoustic signal, which a cloned voice cannot fully hide, rather than to how plausible the request sounds.
- 02Run detection where banks take instructions by phone: the relationship manager's line, the IVR and the agent desktop.
- 03Confirm large instructions through a channel the bank chose, not one the caller or their lawyer supplied.
- 04Keep the verdict with the transaction record, so an investigation starts with evidence rather than recollection.
Hear what the manager could not
Raid AI scores cloned voices on live calls, including 8 kHz phone audio, and in recordings after the fact. Book a demo to test it on your own calls.
