This...sounds incredibly naive? They seem to think that AI risk comes from Bad People doing AI? There's not one mention given to the possibility of well-intentioned people destroying the world by accident.