Dan H comments on Towards more cooperative AI safety strategies