No, quite the contrary. Machines' inability to break laws on their own free will and discretion (because of the lack thereof) doesn't make them more trustworthy, since in any case you would have to trust people who programmed them and which also will be able to hack it as well.
A machine can go wild and this would be even more lethal that a human going mad.
Have you ever seen a machine go wild?
Car makers recall their vehicles because of malfunctioning on a rather regular basis. Some of these malfunctions can actually be lethal, e.g. the unintended acceleration of the Audi 5000 model which was linked to 6 deaths and approximately 700 accidents in 1982-1987.
Those machines were broken or defective. They didn't go wild. Besides, they all had safety mechanisms the operator failed to utilize such as neutral gear and brakes.
It was a figurative expression for being broken or defective to give a tint of rationality, or, rather, ability to lose one (as in "lose control"). I think you got what I meant to say, since here we are all endowing machines with human qualities (and no, rationality is not a machine quality, in any sense of the phrase).