Ten recommendations to make AI safe for humanity