HelenWatson comments on A general model of safety-oriented AI development