Security Issues for AI

Image source: https://ai.googleblog.com/2018/09/introducing-unrestricted-adversarial.html