Many countries are extensively implementing face recognition video surveillance systems. Such a measure, in their opinion, can protect the public from crime, but many civil society representatives see them as a threat to their freedoms and are looking for ways to bypass the ubiquitous tracking system.
Researchers from the Belgian K.U. Leuven University have published an article featuring the ways to bypass image identification systems. For example, facial recognition sensors can be fooled simply by hanging a colour printout around your neck that would make you invisible to artificial intelligence
The fact is that any AI existing today has its flaws, using which you can hide from their all-seeing eyes. Even minor changes in the picture can break the machine learning system resulting in the robot seeing things that are not there. Such tricky images are called adversarial examples.
The use of adversarial examples definitely involves certain dangers. For instance, you can become invisible to an unmanned vehicle and cause an accident.
Share this with your friends!