And, be fair: there's probably a sensible kernel to this rule; to wit, just because you see something on the news doesn't mean it's OK for you to talk about: (1) by doing so, you could be confirming facts that are mere speculation, (2) you run the risk of disclosing more than was actually disclosed, (3) you inherently disclose "metadata" about how widespread the information was, (4) you make it that much harder for the genie to be put back in the bottle in cases where that's feasible (admittedly, the Internet seems to moot [4]).

Like I said, the rule is probably moot in Wikileaks case, but probably not moot in others. Meanwhile, the classification system does revolve around substituting rigid rules for individual judgement --- and that makes sense, given the sheer number of people who come into contact with CI.