There’s no denying that in recent years this technology has caused a lot of confusion, manipulating public opinion through false videos or photos. But in the end we managed to find an effective solution to fight it.It looks that way, because that successfully developed an experimental neural network that is able to efficiently distinguish the originals from DeepFakes. All thanks to researchers at the University of California, who started their work from a vast collection of original and manipulated images, marked according to artificial intelligence have a foothold. On the converted images indicated pixels around the boundaries of objects digitally added, because as previously able to determine these are slightly different than in the case of components originally present in the shots.Of course, the human eye is unable to detect this subtle change, but the computer analysis pixel by pixel so already. Yes described DeepFake photos were then made available to the neural network uses algorithms modeled on the human brain and designed to capture certain patterns in the analyzed database. Based on these images AI has learned to identify previously mentioned limits of what has been proven by the work of the pictures, which we had not seen before. Of course, as of now we are not talking yet about 100% efficacy and say the researchers, it may not even be possible, but ensure that their solution can cope perfectly in “most cases”. Moreover, the effectiveness of the current level is sufficient to pre-filter the content, which are then reviewed by humans. What’s more, the neural network is currently only works in the pictures, but the researchers are already planning similar tests in the case of video, where it will work exactly the same way, working on individual cages.As the authors of the new solution: – It is possible that the man was able to see everything that is currently on the network, but on the other hand, do not do this reliably the machine itself. It must be a combination of both. Is it really a new neural network will help us in the fight against DeepFake? It is difficult to predict, especially that so far there is no question about it yet to a particular entity was interested in their rapid implementation, but there is no denying that DeepFake a big problem and something modern network in the end you will have to do with it.