Thanks to artificial intelligence, everybody can easily create deepfakes without any particularly technic knowledge. By analyzing faces movements. But these will become more and more realistic as technological developments progress, and therefore more and more problematic ...The rapid evolution in synthetic image generation and manipulation has now come to a point where it raises significant concerns on the implication on the society. With the advent of fake news and its harmful effects on social networks, the dissemination of deepfakes on the web therefore constitutes a new technological threat. Manipulation, disinformation, humiliation, defamation ... the dangers of deepfakes will be more and more numerous. In this post, first, we will describe in brief how deep learning with Depthwise Separable Convolutions can be the most useful and promising techniques to detect deepfakes