Generally speaking, as I said, intelligence agencies tend to abide by this notion of an implicit caveat. There are also explicit caveats that are placed on intelligence that will say “Not to be used for...”, and you fill in the blanks. We rely upon our partners to follow those caveats, as they rely upon us to follow theirs. There is a mutuality to it.
I think that when you step out of the Five Eyes, European intelligence agencies will probably follow the caveats as well, for the most part. When you get further afield, it's much more difficult to ensure, but within the Five Eyes I think caveats are well respected, and everybody gets the protocol.
I want to add something very briefly. I'm going to part company with my colleague here on the question of retention, and I'm going to do it in this way.
To the extent that the service properly receives—and I underscore “properly”—information about national security threat information, I don't think it's wise for the service to destroy it. I don't think it's wise. Today it may not mean much, but 10 years from now or five years from now, when circumstances change and you're continually revising your analytics, you need to have a rich environment to be able to stay on top of the threat environment.