Nathan Helm-Burger comments on Alignment Faking in Large Language Models