Oliver Sourbut comments on Confusions in My Model of AI Risk