Mark Xu comments on A space of proposals for building safe advanced AI