My biggest pet peeve with this work is that they incorrectly say: > These adversarial examples have been recognized since 2014 (here's one of the first papers

My biggest pet peeve with this work is that they incorrectly say: > These adversarial examples have been recognized since 2014 (here's one of the first papers on the topic). Adversarial machine leaning has been around since 2006. It's only since 2014 that people decided to call them Adversarial Examples.