Krieger comments on my current outlook on AI risk mitigation