interstice comments on No Safe AI and Creating Optionality