Company Scammed Out of $25M Using Deepfake Technology

( – The advance of AI and related technologies just cost one Hong Kong firm $25 million after an executive was misled during a virtual meeting by lifelike simulations of his coworkers.

Deepfake technology allows computers to be trained on input data such as voice recordings or video of their faces, and to extrapolate the basic components like tone, timbre, and head movements. Software can then be used to make an accurate facsimile of the individual being deepfaked.

Scientists have demonstrated the potential in legitimate experiments, but criminals have begun to deploy the technology in a variety of schemes. One such scheme edited a clip of the JRE podcast to create the impression Rogan was endorsing a male enhancement product.

Deepfakes themselves aren’t illegal, and many exist online as parodies. When properly identified as such, they’re harmless. The problem arises when these convincing simulations are used for criminal fraud.

WIO News reported the first-of-its-kind deepfake attack where criminal actors isolated a specific individual. Using faked messages from multiple “coworkers” they convinced him to make a secret money transfer. The police reported that the scammers used the lifelike simulations to order multiple transfers to different bank accounts, but didn’t actually interact with the individual in any meaningful back-and-forth interaction. This suggests the tech is still early in development.

Previous attacks have used one-on-one simulations, wherein the victim was convinced by a single faked individual. This marks the first time attackers were able to simultaneously create multiple fake personas to mislead their target.

Hong Kong police have already made 6 arrests in the case. They’ve said there have been at least 20 such incidents across a variety of businesses. Deepfakes have been used to open bank accounts, attempt to secure loans, and other financial malfeasance using identities that have been stolen.

The employee began to make inquiries amongst his actual coworkers, discovering that the incident had been entirely engineered. Once he realized the issue, it was immediately reported to the authorities.

Copyright 2024,