Shmi comments on AI risk, new executive summary