As humans, robot’s commanders are not infallible. In this Guest Post Matthias Scheutz argues that to truly act in their inventor’s interest, robots need to be able to disobey instructions.