A fake 3D head model fooled Android's facial recognition. Apple's Face ID was the only one that resisted

Magazine Forbes published a rather interesting test a few days ago, the aim of which was to demonstrate the level of security of mobile authorization systems that use facial recognition elements. To bypass the security mechanisms, a relatively detailed model of a human head was used, which was created with the help of a 3D scan of a person. Systems on the Android platform flopped, while Face ID, on the other hand, did very well.

The test pitted top models from several smartphone manufacturers against each other, namely the iPhone X, Samsung Galaxy S9, Samsung Galaxy Note 8, LG G7 ThinQ and One Plus 6. A 3D head model, specially made after a 360-degree scan by the editor, was used to unlock it. This is a relatively successful replica, the production of which cost over 300 pounds (approx. 8.-).

During phone setup, the editor's head was scanned, which served as the default data source for upcoming authorizations. The testing then took place by scanning the model head and waiting to see if the phones evaluated the model head as the "message" and then unlocked the phone.

In the case of Android phones, the artificially created head was 100% successful. The security systems in the phones assumed that it was the owner and unlocked the phone. However, the iPhone remained locked because Face ID did not evaluate the head model as an authorized target.

However, the results were not as clear-cut as it might seem at first. First of all, it should be mentioned that other manufacturers warn that their facial scanning phone unlocking system may not be 100% secure. In the case of LG, there was a gradual improvement in results during the test as the system "learned". Even so, the phone was unlocked.

However, once again, Apple has proven to have top-notch facial scanning technology. The combination of infrared object meshing and creating a three-dimensional face map is very reliable. Much more reliable than more common systems based only on comparing two images (model and actual). Another indication of the great functioning of Face ID is also the absence of reports of this system being hacked and misused. Yes, Face ID has already been fooled in laboratory conditions, but the methods used were even more expensive and complex than in the test mentioned above.