“Uses cautionary language ('threatens,' 'improper,' 'might not work') but addresses policy feasibility, not human harm or crisis.”