The problem being that, in both the DC and Marvel universes, robots who are programmed to be irredeemable automatons not only CAN break such programming and gain redemption, but DO so at an alarming rate of frequency. Indeed, if anything, it's The Law Of Contrarian Robot Programming at work, as seen in nearly all sci-fi:
1. If a robot is programmed to be perfectly safe and beneficial to all people, it will inevitably go nuts and start slaughtering people .
2. If a robot is programmed to decive, betray, wound or murder, it will be inspired by the innate goodness of humanity to shed a tear and choose a life of heroism.
Clearly, then, what robot manufacturers should be doing, to create a perfectly safe population of robots, is to program them all to kill all humans, because whatever they're programmed to do, they'll always wind up doing the opposite in the end.