Can Phones Be Unlocked With Deepfake Videos?

The new advantages of technology in bad hands Unfortunately, we can see that it can also be functional. Hacking bots slowing down pages, fake and auto-managed (bot) accounts pretending to be real individuals on social media, and potential to spark political debates deepfake videos We can give an example of this.

In particular, it is necessary to open parenthesis on the deepfake side here. Because the technology in question is more and more fun Although it is used in the field, it is certain that it will reach a quality that cannot be distinguished from the real at some point. Well, this technology, which has the potential to deceive people, provides privacy with facial recognition. our smartphones Can it be wrong too?

The answer is hidden in the working logic of face recognition technology:

The two most well-known phone brands apple and of Samsung We can simply shed light on this issue by looking at facial recognition technologies. Let’s start with Samsung:

Initially, 2D face recognition was used in Samsung and other android models. So what does this mean?

2D face recognition scans your face but does not analyze your face shape as it does in 3D. In other words, both the face you introduce from your camera and the photo you show to the camera. in two dimensions saved. That’s why face recognition systems on phones have only been used in the past. with photos We saw that it was possible. Because only the front camera was used to store your face information. In addition, with machine learning, this system was improving as you used it.

But later on, Samsung realized the danger and started to sell to its flagship models. infrared LED placed it. In this context, the 3D system was adopted and higher security was achieved. But Samsung does this not facial recognition, iris recognition as revealed. This feature, which we saw on the Galaxy S8 models, was unplugged after a while, and in addition to the simple face recognition system we mentioned, a fingerprint option was offered.

“How is 3D scanning different from the other?” If you ask, the infrared LED reflects a large number of dots on your face. 3D face map and thus it is understood that what is shown to the camera when unlocking is not a photo or video.

This is why we hear less about ‘photo unlocking’ over time; Our move to infrared sensors:

However, seeing two-dimensional facial recognition systems that do not provide much security in most models. possible. Machine learning also provides their security.

So what did Apple do?

Apple makes its revolutionary move for iPhone history iPhone X made with the model. Touch ID disappeared when the button that everyone was used to for years was removed from the model in question. Instead, it works with the help of an infrared sensor. Face ID came into our lives. The technology here is as we mentioned in the Samsung section. to infrared projection based on. Here again, about 30 thousand tiny dots are projected onto your face and security is provided by creating a face map in the same way. Besides Face IDIt also adapts to factors that make your face a bit different, such as beard and glasses.

So how can we connect all this to deepfake?

The difference between two-dimensional and three-dimensional face scanning we’re talking about may be a bit confusing when it comes to deepfake, but the situation is actually as in photography. Although you create realistic videos of someone else that look like they are in 3D, the camera will perceive it as two-dimensional as it will see it from the screen. So this is the case with sensors commonly used in the smartphone world.

In this context, the image that the infrared light points will create from the 3D deepfake video will be as follows.

tangibleDeepfake will not be a formidable competitor for infrared technologies like Face ID, unless realistic three-dimensional facial models are obtained, which would be terrifying. Thanks to this security, even shopping can be done with face recognition today.

Do you think this will become surmountable for malicious people in the future? You can share your thoughts in the comment section. In addition, let’s not finish without mentioning that different companies follow different methods for face recognition, but these two are enough for deepfake.