Ebenezer Dukakis comments on Alignment Faking in Large Language Models