New ways to avoid AI surveillance.

Could be usefull once integrated in to a (less flashy) camouflage pattern.

More info in this paper from KU Leuven : Fooling automated surveillance cameras: adversarial patches to attack person detection

Tesla develops the hardware for self driving

Important takeaways:
– Tesla considers LIDAR as a bad technology for self driving
– Tesla cars and the required neural networks for self driving are becoming very good because they collect a lot of data and can retrain the networks for all kind of exceptions. This is for Defence a enormous problem because we operate in a non structured world where very little structural data acquisition is being done… so even with the best hardware, we probably won’t be able to have self driving vehicles that can operate cross country and in the unusual conditions we find in operations.

EEAS PoV on AI for lethal weapons