http://www.theverge.com/2017/4/12/15271874/ai-adversarial-images-fooling-attacks-artificial-intelligence
a good read with examples of camo in the new digital world of Ai.
more up to date examples.
That's an interesting article, and it shows why I prefer to create my own algorithms.
What this shows is that knowing the algorithm used we know how to fool it. :)