ADVERSARIAL ARTIFICIAL INTELLIGENCE

 A student at MIT developed an algorithm that trick AI to misidentify objects. For example, the algorithm can wrongly identify a headphone as a gun at an airport terminal. By switching a few pixels or adding noise to the image, AI can misidentify a gray tabby cat as a bowl of guacamole. 

Resource: https://www.theregister.com/2017/11/06/mit_fooling_ai/

Solution: https://wmx-api-production.s3.amazonaws.com/courses/5741/supplementary/us-18-Parikh-Protecting-the-Protector-Hardening-Machine-Learning-Defenses-Against-Adversarial-Attacks.pdf