I always had the thought in the back of my mind "what better way to flag content as potentially incriminating then to let the user tell you "this conversation is not to be retained."

Especially the way the "advertise" incognito as some sort of untraceable means of searching/chattig etc, we ought to see these sorts of things from a less trusting-by-default perspective.