“Uses 'catastrophe' and 'dangers' to frame AI risks as urgent threats requiring immediate global intervention.”