P. João comments on Alignment Faking in Large Language Models