The dangers of face processing with AI

avatar
Greetings dear friends.

Perhaps it is something that we do not take into account very much, but the processing of faces by software based on Artificial Intelligence is reaching unsuspected levels.

IA rostros 1.png
My appearance manipulated with FaceApps.

On the one hand there are many photo Apps that through the use of AI process and modify our face with an endless number of filters so that we can share incredible photos on social networks, sometimes incorporating funny elements to the face and in other cases modifying the appearance; in the image above we can see how I have altered my appearance with FaceApps, an application that works with AI. But although this may seem innocent to us, we must keep in mind that biometric elements identify us as a person, so we are entering a world where this technology can be used so that we cannot distinguish between what is real and what is fictitious.

Although in fact, this is already a reality, in the following image we can see faces of people that do not exist, they were created by an Artificial Intelligence algorithm called Generative Confrontation Networks (GAN) and that we can find on the web "this person does not exist", very apt name by the way, and the algorithm works so fast that we should not do anything, with visiting the page shows us a face, and reloading the page the algorithm shows a different one created instantly.

IA rostros 2.png
Images generated by the GAN algorithm, taken from thispersondoesnotexist.com.

The author of the algorithm explains that the page tries to demonstrate how easy it is to generate a face with this software, something that could give more realism to video games and other types of animations without the need to resort to greater graphic power; but the dark side of this technology is that it could be used for more dangerous purposes, as it would be very easy for a person to create a profile in a social network very convincing but completely false to attract users for some hidden purpose.

And not only images can be altered, we have already witnessed how, through the use of image processing software, videos can be manipulated to alter the face of a person and place the face of another person who did not appear in the scene. Perhaps many remember that the movie Fast and Furius 7 was finished filming without the actor Paul Walker who died in a tragic accident, some scenes were recorded with his brothers and with the help of technology they altered their faces to make them more similar to the late actor and complete the movie.

Although the above was very emotional for the public and was received almost as a tribute to the actor, the truth is that it can also be used to ruin the reputation of a person, through a technique known as deepfake, which consists of synthesizing images on existing videos and passing them off as real, this has been done to circulate videos where a personality is shown in situations that discredit it. Some of the most notorious cases of deepfake include algorithms that produced pornographic videos with the image of some celebrities, such as Gal Gadot or Scalett Johansson.

Another danger of this technique is that the credibility of a person can be used to spread a false message, as demonstrated by film director Jordan Peele when he produced a video showing former President Barack Obama calling Donald Trump a dipshit; the director claims to have made the video to warn about the potential danger of Artificial Intelligence to spread Fake News.


Deepfake of President Obama performed with AI. Source: BuzzFeedVideo

These videos where the face and voice of a person is manipulated with Artificial Intelligence programs also represent a dangerous form of fraud, since they can call a businessman by phone with any excuse and record his voice, then they can synthesize his image and impersonate his voice in order to sell a product in his name or call one of his employees and order him to make a transfer or request information from the company.

Since the beginning of the century we live in the age of information and communication, but we are entering a time when a person can create a very real fictional character or impersonate another to commit some scam or spread a false message and confuse the population about a certain fact, it is clear that in the future we will need legal limits regarding what AI can perform regarding face manipulation.


Thanks for coming by to read friends, I hope you liked the information. See you next time.




0
0
0.000
2 comments
avatar

Hello, the information you share is very important, because of the negative consequences of using it for fraudulent or criminal purposes.

What may initially be considered fun, has implicitly other consequences of which we must be careful.

Thank you for sharing.

Regards

0
0
0.000
avatar

That's right friend, what for some is just something fun, others will find a way to use it for fraudulent purposes, these are things that we must be aware of and the systems will be looking for ways to contact this type of manipulation.

0
0
0.000