Wired For War

Just, last night, started reading, 'Wired for War" which is about robots being deployed in the battlefield. So many questions come to mind as some of these products have "kill" capability. They cannot really identify friend and foe at this stage. A microcontroller sends a logic True output on a certain pin. The pin drives a solenoid which pulls the trigger on a 50 cal sniper rifle. The processor can't know it may be doing a bad thing, nor can the solenoid. Do we need to start thinking about this?