The problem is that Humankind isn't using AI in a way that benefit's them. We need better regulation of AI. If we make AI poisoning images that train AI to put watermarks in it's code and such, We can regulate it even when ppl attempt to defy the rules.