In the middle of a post on Larry Craig, Mark Schmidt interestingly says “[I]n my world, if something’s none of my business, it’s o.k. for you to lie about it, in order to protect your privacy.” That would allow a much broader sphere of permissible lying than many philosophers would (I think) allow. Still, it sounds like a pretty plausible principle to me.
I'm more sympathetic to this idea now than I was last year -- in part because I'm less inclined to assume that (even intentionally) false statements are properly considered "deceptive". Cf. Nagel: "The point of polite formulae and broad abstentions from expression is to leave a great range of potentially disruptive material unacknowledged and therefore out of play... this is not a form of deception because it is meant to be understood by everyone."
I think it is always (pro tanto) wrong to intentionally deceive others, i.e. to take as your goal the manipulation of their beliefs so as to introduce falsehood. Ideally, we would have social norms such that "lying" to protect privacy is recognized as standard practice, so nobody would actually be deceived. But even failing this, we may introduce a kind of 'doctrine of double effect': if you say something false in order to protect your privacy, which has as an undesired (albeit predictable) side-effect that it gives rise to false beliefs in the listener, then that's morally okay. One did not intend to be deceptive; the listener's beliefs are mere "collateral damage".
One worry here is that the deception is no mere "side effect", but the essential means by which one's goal of privacy protection is achieved. This will depend on the case. I mean to defend those lies that serve a purely deflective purpose (i.e. "where refusing to answer cannot preserve the secret"). In other cases, one might lie for remedial purposes, i.e. to actively undo some damage that has already been done. If your privacy has already been violated, you might wish to manipulate others' beliefs back towards the state they were in before the violation. I'm not sure whether I have any general views on the im/permissibility of this (as opposed to case-by-case judgments), but such cases would count as genuinely deceptive lies, and so be at least pro tanto wrong (even if possibly justified in light of other reasons).