This whole situation shows once again how advanced face fraud technology has come. Scammers have used publicly available videos to create fake interlocutor avatars that probably no one even noticed were fake.

Detail

According to the source, the scammers made numerous video calls. There was probably only one person in the frame at all times whose appearance was altered by the AI. Others in the frame are believed to be entirely digital and generated by the program. Law enforcement officials say they’ve never seen anything like this before. Usually only one person is fake.

Preparations for this operation began in mid-January, when a phishing message was sent from the company’s head office to its Hong Kong branch. It was not explained what was in this message. However, it can be assumed that this is a video conference announcement or a message that a secret money transfer needs to be made.

In short, the finance department employee was deceived into believing that the transaction was necessary. Despite initial concerns, the employee participated in a group video conference that featured fake versions of the company’s CFO as well as other employees who looked and spoke like their real colleagues.

This time, during a video conference with several people, it turns out that not everyone you see is real. They used deepfake technology to impersonate their targets’ voices.
Senior Inspector Chan Shun-Chin told SCMP in an interview.

According to Chan, three employees fell victim to scammers. Another also attended the video conference. They all say the people in the frame “look real.” So they made a transfer of HK$200 million.

Deepfake technology has been around for years. We’ve seen plenty of videos of celebrities saying things they’ve never said, Barack Obama kicking down doors, Elon Musk promoting cryptocurrencies, and more. But such frauds have always been at a fairly low level. Until the moment when modern artificial intelligence comes into play. It allows you to increase the realism of the created image to such an extent that the result is literally indistinguishable from the original. For some neural networks, just a few seconds of video or audio is enough to start mimicking a person’s appearance and voice. Last week, unidentified individuals began mass distributing fake pornographic images of singer Taylor Swift, causing a real uproar in the media and social networks.