Company out $35M after scammers stage video call with deepfake CFO, coworkers

The employee was tricked into transferring $200 million Hong Kong dollars into five bank accounts after joining a video call with convincing AI fakes of their coworkers.

A multinational firm in Hong Kong is out 200 million Hong Kong dollars (around $34.5 million Canadian) after a financial worker at the company was targeted by scammers using artificial intelligence, culminating in a phony video conference call with numerous deepfake colleagues.

Police discussed details of the incident, without naming the company or worker involved, during a press conference in order to warn the public about the novel scam.

Acting senior superintendent Baron Chan of the Hong Kong Police Force’s Cyber Security and Technology Crime Bureau said the scam began last month when the worker received an email, purportedly from the company’s U.K.-based chief financial officer (CFO).

The email was concerning a “secret transaction” that needed to be carried out, according to the South China Morning Post. The employee had an early “moment of doubt,” as the email appeared to be a phishing scam, but they were eventually fooled after the fake CFO invited them to a video conference call.

On the call appeared to be numerous other coworkers that the employee recognized — they even sounded like the real deal. But they weren’t the person’s coworkers; they weren’t people at all. They were deepfakes, a type of synthetic media created through machine learning that can mimic a person’s appearance and speech. A digitally recreated CFO and a few outsiders were also present on the conference call.

“Because the people in the video conference looked like the real people, the (employee)… made 15 transactions as instructed to five local bank accounts, which came to a total of HK$200 million,” Chan said during the press conference, broadcast by Radio Television Hong Kong.

“I believe the fraudster downloaded videos in advance and then used artificial intelligence to add fake voices to use in the video conference,” Chan added.

Experts have warned that AI voice and video generation is becoming easier to access as the technology improves. It used to take extensive recordings to create a believable cloned voice, but now, it only takes seconds of recorded speech. Any public video can feasibly be used to train an AI model to mimic a person’s voice and appearance.