Davidmanheim comments on an effective ai safety initiative