AI ethics - how do we put theory into practice when international approaches vary?

Many of the issues around AI and ethics are understood, but what practical steps should we take to protect citizens?