Protecting humans should be the highest priority of any responsible AI system. At any capacity, AI must not harm humans physically or emotionally, directly or indirectly.