finnhambly comments on A Playbook for AI Risk Reduction (focused on misaligned AI)