A way to fool facial recognition systems

If you don’t want existing cameras all over the world to capture your face and find you on social media, there is a program that can prevent it.

It is the product of a team at the University of Chicago, which has devised a subtle tactic that effectively fights this type of spying algorithm.

The program, called Fawkes, in homage to the Guy Fawkes mask, began rolling out late last year as a way to thwart companies like Clearview AI, which compile databases of faces in public publications. can recognize the appearance of a given person by connecting an image of a face (i.e. from a Facebook profile) to another image of a face (i.e. from a passport photo) and find similarities between the two Photos. According to the Chicago team, this not only means finding matching facial geometry or matching hair color or matching moles, it also means detecting invisible relationships between the pixels that make up a computer-generated image of that face.

What the program does is swap or distort some of these pixels, the face may still be recognizable by people, but not by computers, since they fool the popular facial recognition algorithms.

According to the team’s research, this cloaking technique managed to fool facial recognition systems marketed by Microsoft, Amazon, and Google 100% of the time.

The Fawkes program is freely available for download on their website, so if we have an image that we want to protect from snoopers, we can use it to blend those invisible pixels in about 40 seconds per photo. The resulting photo can be uploaded on the social media platform of our choice, with the guarantee that it will not cross any other published by us.

Back to top button

Ad blocker detected

You must remove the AD BLOCKER to continue using our website THANK YOU