At the dawn of the era of robotization, the outstanding American science fiction writer Isaac Asimov formulated three main laws of robotics, the first of which says: under no circumstances should a robot be capable of harming a person. This law must be fully respected in their training. What happens if non-humans are trained in robots? We are talking about animals that, of course, do not train robots, but if the robots contact them, they will take a completely different attitude towards humans, which can create many problems in the future. As Dr. Hamd and Andrew Hein (California) found out, artificial intelligence is capable of perceiving animal behavior, which can lead to robotic attacks on humans in the future. According to experts, this is due to the ability of robots in the future to perceive knowledge not only from humans, but also from animals. Such an incident has already occurred in Australia: a robot, having spent some time next to wild wolves, began to growl at its creators. How far this process goes, it is difficult to predict, however, due to contact with animals, robots, adapting to the outside world, will be able to show aggression regarding a person. According to Andrew Hayne, at present, the algorithms of artificial intelligence embedded in robots are configured for training, but this does not exclude the possibility that robots will also receive a kind of “training” from other living beings. Japanese scientists have already warned of the need to prevent the possibility of robots attacking humans, but experts believe that the implementation of this concept is possible only by 2020. At the same time, experts already state the fact that the logic by which the neural network makes a decision is no longer always understood by them.