The new frontiers of DeepFake

Better call… Trump? A video showing the US president in the role of Saul Goodman, the protagonist of the TV series “Better Call Saul” is becoming viral. This is not a re-casting but the extraordinary result of a DeepFake algorithm. We have talked about it often, but in this case, we try to make a point about the situation starting from the birth of this technology.

In the beginning, there was research. The technology that today we call DeepFake has deep roots, already in the late 90s many researchers worked in the field that was called “computer vision”, that is the field that tried to allow computers to process images and videos. The first concrete result was the project called “Video Rewrite” which, as early as 1997, made it possible to modify the movement of a person’s mouth to make it appear to say other words. We will have to wait 20 years in which this research undergrowth has fermented to obtain a credible version in the form of a video, created through machine learning algorithms, of Obama in which the movement of the mouth and the audio are taken from another speech of his.

The DeepFake bomb. In the same year the term DeepFake was born, not from a scientific publication but thanks to a user of the Reddit site called DeepFake (a name that derives from the union of Deep Learning, a machine learning method that allows the generation of this content and Fake, fake) who began to publish on the platform several pornographic videos in which the face of the protagonists had been replaced with that of famous actresses through this algorithm. In this period we are witnessing the first public discussions on this technique and its ethical implications.

The ban of DeepFake. Following the grievances of public opinion and the actresses involved in the affair, many sites, social networks and even pornographic sites like PornHub have banned these videos as they are considered a form of involuntary pornography. Not all videos produced with DeepFake techniques, however, concern pornographic content, many are created, as in the case of the Trump video above, for pleasure and in search of virality.

Worries. With the improvement of this technology, we are approaching a period in which to create DeepFake are no longer necessary computer skills and cutting-edge equipment, indeed, already make peeping app for smartphones capable of inserting our face into movies and TV series. Precisely this ease of creating content DeepFake is raising many concerns: systems of this type would easily lend themselves to creating highly credible hoaxes in all areas, making statements unthinkable to political leaders or else the abuse could go so far as to touch the revenge porn.

The future of DeepFake is uncertain on which it is fundamental to discuss, on the one hand, they demonstrate the incredible skills that we have been able to achieve with machine learning techniques and could also help save lives, on the other it is a powerful technology that is bad hands can also lead to serious consequences.