On a side note and /rant for AI guidelines just follow Asimov's Laws of Robotics. Throw in the beginning rules for the Hippocratic Oath and you can effectively call the rules done. "Do no harm" Is a pretty strict rule on top of Asimov.