LLMs don’t reason and they don’t understand anything, therefore there is no way LLMs could apply any rules based on abstract concepts such as “harm” or even “human.”

It would make as much sense to ask my dog to govern himself according to French common law.