Home » Technology » Innovation & Startup: Facebook is developing a system to fight against deepfakes

Innovation & Startup: Facebook is developing a system to fight against deepfakes

You know, for several months now Facebook has been trying to fight against fakes news and in particular deep fakes. And it’s nowhere near as easy as you might think. This is why, Facebook researchers have teamed up with scientists at the University of Michigan to develop a new system to detect deep fakes. The system appears to be very efficient.

An interesting novelty

On June 16, Facebook officially broke the silence to announce its new collaboration with the University of Michigan. This collaboration made it possible to develop an artificial intelligence not quite like the others. The goal is to detect deep fakes and go back to their source to find out who created them.

You should know that deep fakes are fake images that are produced to manipulate Internet users and to propagate fakes news. This is a real problem for the social network which has been seeking for several months now to fight against those who are at the origin of this misleading content. To fight against deep fakes, Facebook has decided to use artificial intelligence which makes it possible to disentangle the true from the false quite easily.

An effective solution

Facebook has already proposed several solutions to detect deep fakes. Unfortunately, the results did not live up to expectations. However, it seems that the artificial intelligence developed is able to detect deep fakes very effectively. We are talking exactly about reverse engineering. That is, artificial intelligence studies the system to understand how it works. She will deconstruct the photo to identify any intersections that may have been added during editing.

Thus, artificial intelligence highlights fingerprints that make it possible to identify with certainty an image that is not true. In addition, artificial intelligence will cross the similarities in the creation of these photos and images to go back to the source. For now, the project is still in its early stages, but it looks particularly promising. Case to follow


Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.