The algorithms that computers use to determine what objects are–a cat, a dog, or a toaster, for instance–have a vulnerability. This vulnerability is called an adversarial example. It’s an image or ...