Deepfake is the name given to the technique that allows to recreate the human appearance or voice with the aid of artificial intelligence. With technology, it is possible to simulate people speaking or acting in ways they never would.
The word would be a junction of the terms deep learning (a branch of machine learning or machine learning, in Portuguese) and fake (false). The concept would have appeared in 2017, in reference to a Reddit user with the nickname of deepfakes.
He would have used the technology of deep learning to overlay the faces of celebrities and those of people in pornographic films. Since then, the method has become increasingly realistic and accessible.
AppGeek has prepared a guide to better understand what it is and how the editing feature that has been dominating the internet works. Also check out what are the risks offered and how to protect yourself.
How does deepfake?
Actress Michelle Adams has her face replaced by that of actor Nicolas Cage (Reproductionb / Wikipedia)To create a video deepfake two competing artificial intelligence algorithms, called generator and discriminator. According to the Whatis.com website, the first job is to create fake videos and, second, to identify them as true or not.
When the discriminator finds false content, the generator learns about what should not be done. And so, you will create the next video in an improved way. When they act together, generator and discriminator generate the call generative adversarial network (GAN), or generative opposition network, in free translation.
How to create a video deepfake?
To assemble a film of the genre it is necessary to have a source video, preferably as simple as possible. For this reason, political speeches are often used, in which the background is neutral and basically head movements are made.
So, you need to combine two sets of data. If the character in the video turns his head to the left and opens his mouth, it is necessary to have images of who will replace him in the same positions. This may require a vast image bank.
Then it is up to the artificial intelligence-based program to leave the job. After several trials and errors and successes, the generative adversarial network will incorporate one face to the other as accurately as possible.
The video above is one of the most famous deepfake. In it, former President of the United States Barack Obama has movements and voice overlapped by those of comedian Jordan Peele.
What are the risks of deepfake?
Although it seems harmless, the deepfake has been controversial since its appearance in 2017 on Reddit. The technique was widely used to superimpose the face of actresses on that of people in porn videos.
As it develops, the technique becomes more realistic and the contents end up going viral as if they were real. This is one of the great dangers of the resource: the spread of fake as authentic content.
In 2018, Facebook found itself involved in a scandal following the leak of user data by a third-party app company. Shortly after, a video by Mark Zuckerberg went viral and worried a lot of people.
In the short film, Facebook’s CEO insinuates that he controls the future because he has the domain of billions of data. The content was nothing more than product deepfake created by Bill Poster for Specter.
Another predicted problem is the loss of credibility of any type of material published on the internet. Technology makes it increasingly difficult to identify and filter what kind of news is actually or simply created by the computer. What your eyes and ears see and hear may be just a software product.
With the popularization of programs that allow laypeople to create videos of the genre, it is possible that ordinary people are also increasingly targets of this type of montage.
The results can range from simulating participation in adult films to creating false evidence against someone.
As a recent phenomenon, governments do not yet have specific laws to combat crimes related to deepfake. In the UK, the producer of content of the genre may be indicted for harassment.
In the United States, the charges can range from identity theft, cyber stalking and revenge pornography. In Brazil, there is still no specific regulation on the subject.
How to identify one deepfake?
Despite the evolution of the technique, a good part of deepfakes amateurs can still be identified with the naked eye. In such cases, it is possible to identify unnatural blinking of the eyes, movement of the lips or the presence of shadows in the wrong places.
However, the concern about the future is real. In the United States, the Defense Advanced Research Projects Agency (DARPA) is already investing in ways to better identify videos.
For the time being, the best measure is to always be alert to what you receive and be aware of the sources of sharing.
What is ZAO?
Announced in September 2019 on the Chinese App Store, the app ZAO made news all over the world. The program went viral for taking smartphones to the creation of deepfakes from a photo in just 8 seconds.
The result is close to perfection and was shared on Twitter by the user @AllanXia. The face of actor Leonardo DiCaprio has been replaced by his in several scenes and the expressiveness is impressive.
In case you haven’t heard, #ZAO is a Chinese app which completely blew up since Friday. Best application of ‘Deepfake’-style AI facial replacement I’ve ever seen. Here’s an example of me as DiCaprio (generated in under 8 secs from that one photo in the thumbnail) ? pic.twitter.com/1RpnJJ3wgT
– Allan Xia (@AllanXia) September 1, 2019
The launch of the app reopened the debate on technology and privacy. Firstly, according to the terms of use of the program, when using it, you would be giving the company the right to use your data and face for commercial uses in a free and irrevocable way.
After many complaints, the responsible company changed the term. Now, he says that the content will be used only by developers for the purpose of improving the application and the technology used.
Another issue of concern is the fact that the app makes it even easier to create overlays with images from anyone. And with an ever greater realism. This could make it easier for victims of fake videos to be viewed as real.
Other apps by deepfake
But this is not the first time that deepfake go viral. The first was FakeApp, released in January for 2018 for desktop. It made it possible to create and share videos with replaced faces relatively easily.
The main difficulty was to gather a large amount of visual material from the person to be inserted, so that the result would be satisfactory. Currently, it is no longer possible to find the available program.
Interested in apps that allow you to change your look? Then meet Snapchat. The famous application allows you to change faces with friends and apply filters to your face in photos and videos in real time.
Also check out 7 apps to turn your photos into drawings easily and 8 super fun apps to change your voice.
