Close ad

As a biometric authentication mechanism, more and more smartphone manufacturers are relying on facial recognition. Abroad, payments in shops, ticket purchases in public transport are even approved with the face, or passengers themselves check in at airports after scanning the face. But as research by artificial intelligence company Kneron shows, facial recognition methods are vulnerable and relatively easy to circumvent. One of the few exceptions is Apple's Face ID.

In order to analyze the level of security of available facial recognition mechanisms, researchers from the American company Kneron created a high-quality 3D face mask. Using it, they managed to fool AliPay and WeChat payment systems, where they were able to pay for the purchase, even though the attached face was not a real person. In Asia, facial recognition technology is already widespread and is commonly used to approve transactions (similar to our PIN, for example). In theory, it would be possible to create a mask of the face of any person - for example, a famous person - and pay for purchases from their bank account.

3D Face ID mask

But the results of tests carried out on mass transport systems were alarming. At the main airport in Amsterdam, Kneron managed to fool the self-check-in terminal with just a photo displayed on the phone screen. In China, the team was able to pay for a train ticket in the same way. So, if anyone wanted to impersonate someone else while traveling or pay for a ticket from someone else's account, all they would need to do is a publicly available photo downloaded from social networks.

However, Kneron's research also has positive results, especially for Apple users. Even a relatively believable 3D mask, the creation of which was expensive and time-consuming, could not fool Face ID in the iPhone and iPad. The facial recognition mechanism in Huawei's flagship phones also resisted. Both systems do not rely only on the camera, but capture the face in a more sophisticated way using infrared light.

Source: Frotune

.