Intelligence Sharing Runs on Trust That Is Wearing Thin

Intelligence liaison is not a treaty relationship. It is a set of habits between services, sustained by individual officers who trust their counterparts and by an institutional expectation that shared material will be handled a particular way.

The originator control principle is the load bearing element. A service shares on the understanding that the recipient will not pass the material onward, will not act on it in a way that exposes the source, and will not use it in a domestic political argument.

Each of those has been tested in recent years, in more than one country, and services respond to the tests the way any careful institution would. They share less, they compartment more tightly, and they route around a partner rather than confronting it.

The degradation is invisible from outside, which is what makes it dangerous. Nobody announces a reduction in sharing. The material simply stops arriving, and the gap is discovered later, usually during something that would have been easier to see coming.

The absence of a treaty is the feature, not an oversight. A liaison relationship that required ratification would be slower to form, harder to adjust and impossible to end quietly, and none of those properties is wanted by either side.

What substitutes for the treaty is a set of understandings, and the load bearing one is originator control. The service that collected the material decides where it may go, and the recipient's compliance is voluntary in the sense that nothing enforces it.

Nothing except the obvious: a service that breaks the understanding stops receiving. That is the entire enforcement mechanism, and it is usually sufficient because the value of continued access exceeds the value of any single disclosure.

The domestic legal frame sits underneath without governing the relationship. The classification order determines how received material is handled once inside, and the statutory framework sets what the receiving service may lawfully do with it.

Those rules protect the material and say nothing about the confidence, which is the thing that actually determines what arrives. A partner assessing whether to share is not reading the recipient's classification order.

They are assessing two risks, and both have been tested. The first is leakage, whether through compromise or through a domestic disclosure process the partner does not control.

Disclosure law is the visible half of that risk and the more manageable one, because its exemptions are well established and a partner can predict how they operate.

The second risk is political use, and it is the corrosive one. Material shared for a security purpose that surfaces in a domestic argument tells the originating service that its product is an input to someone else's politics.

A service that concludes this will not announce a reduction. It will share later, share less, add caveats, or route around the relationship entirely, and each of those is deniable as ordinary caution.

That invisibility is what makes the damage hard to govern. The oversight office can report on the classification system in aggregate and cannot report on what a foreign partner has quietly stopped sending.

The gap surfaces only when something is missed, and by then the causal chain runs back years through decisions nobody recorded as decisions.

Which is the argument for treating these relationships as infrastructure rather than as a resource. They are expensive to build, invisible while working, and the interval between damaging one and discovering the damage is measured in political cycles.

It is worth resisting the assumption that more sharing is always better, because it is not. A service that passes everything upward degrades its own judgement, and a recipient drowning in unassessed material is not better informed.

The thing worth protecting is narrower. It is the confidence that allows a partner to share the small number of items that matter most, which is the first thing lost and the last thing rebuilt.