This is spot on:-
--"Kingston in his paper explored the issue of accountability for AI, which he defines as any system that can recognize a situation or event and then take action through an IF-THEN conditional statement.
Yes, it's a low bar. It's pretty much any software, and yet it's arguably one of the better definitions of AI because it avoids fruitless attempts to distinguish between what is and isn't intelligence." --
Any software can convert the system to the purposes of its designers, rare is it that a system will understand and limit the behavior of software or AI, it would have to be a better System AI than the AI being run on it.
#Criminal Insanity. Is where a Group, Corporation or Body (and/or ...a.System) have no mind of their own and act Negligently and Incompetently, and hence are deemed Criminally Insane. (at least, in Australia - something like this anyway.) [Criminal Insanity NOT the Insanity Defense].
Requirement therefor - The construction and deployment of 'any device' with sufficient rigor to determine that not harm will come from operation, or from anticipated errors and failings.
A Device is a mechanism, system, behavior, figure of speech, scheme, design, tool - and One who employs devices = deviser -> intention
AI should always be supervised by a capable adult or two.