Deep learning for password cracking

Modern password cracking tools, such as John the Ripper, allow a hacker to test billions of potential passwords in a matter of seconds. Not only do such tools allow a hacker to try out every password in a dictionary of common passwords, but they can also automatically transform these passwords by using concatenation (for example, password1234), leetspeak (p4s5w0rd), and other promising techniques. Though these techniques are promising, finding additional promising transformations is a difficult task. The ML system known as PassGAN uses a generative adversarial network (GAN) to automatically learn such rules by observing large datasets of real passwords (gathered from a corpus of actual password leaks) and to generate high-probability password candidates. In this recipe, you will train PassGAN on a corpus of leaked passwords and use it to generate...