The Microsoft co-founder and billionaire philanthropist told Reuters in an interview that he believes China might agree to restricting potentially dangerous AI model releases if the U.S. took initiative on that first.