thomblake comments on against “AI risk”