AI attacks, and especially adversarial ML is an evergrowing threat that has finally been given an appropriate threat model. However, the field is still deeply lacking in defensive theory. The goal of the talk is to go over the current threats posed by Adversarial ML and other AI attacks and discuss the proposed theories for fixing these problems. The goal of the talk will be to make sure pentesters know what to look for when attacking AI solutions, and that defenders know what to consider when protecting these systems.