In gist, if your ingroup does things that harm others, you are likely to subsequently shift your moral attitudes away from principles that tell you that harming others is wrong, and towards principles that value loyalty and obedience.

A more generalized version of this would read: "if your ingroup does [x], you are likely to subsequently shift your moral attitudes away from principles that tell you that [x is bad], and towards principles that [tell you that x is good as long as it's your ingroup doing it]." The chapter's from Cialdini's Influence social proof and identity self-modification seem relevent.