loup-vaillant comments on against “AI risk”