The Pentagon's Public Disclosures Need a Mosaic Test

Defense doctrine already recognizes that individually harmless facts can become sensitive when assembled. Public campaign communications should apply the same logic without retreating into broader secrecy.

Department of Defense classification guidance explicitly accounts for classification by compilation. Its own marking manual illustrates how pieces of unclassified information can become classified when combined because the compilation reveals something more sensitive than any individual piece. The Operations Security program makes the same point from another direction: an adversary can interpret and piece together open-source indicators to estimate friendly intentions, capabilities, or activities.

That logic should be made a routine part of public campaign communications. Call it a mosaic test.

A safe tile can still change the picture.

Operation Epic Fury shows how quickly a modern campaign can produce a rich public record. During the operation, the Pentagon publicly reported that more than 50,000 service members were supporting the campaign and that U.S. forces had struck roughly 6,000 targets by mid-March. Later briefings described more than 10,000 missions and additional campaign-level results.

Those disclosures can be entirely appropriate and still illustrate the analytical problem. An adversary does not read a force count as a press-release writer does. It can compare that count with deployment announcements, public imagery, contract data, replenishment schedules, maintenance reporting, budget documents, congressional testimony, and previous statements. The value may emerge from the relationship among the facts rather than from any single fact.

The Pentagon’s own OPSEC framework is built around this idea. An indicator is not necessarily a secret. It is a detectable action or piece of open-source information that can be interpreted in context. The security question is therefore not simply, “Is this number classified?” It is also, “What estimate becomes easier once this number is added to everything already public?”

That distinction matters because modern open-source analysis has become faster, cheaper, and more automated. Commercial imagery, searchable hearing transcripts, procurement databases, social media, shipping and aviation data, archived press briefings, and machine-assisted extraction can turn scattered releases into a structured dataset. The collection burden that once limited mosaic analysis is falling.

The answer should not be to classify routine information merely because software can aggregate it. That would sacrifice speed, trust, and useful transparency. It would also create more protected material without necessarily protecting the information that matters most.

A practical mosaic test could be narrow enough to avoid slowing ordinary public affairs.

First, high-value operational releases should be reviewed against a rolling disclosure ledger rather than only against the draft in front of the reviewer. The ledger would track recently released force levels, readiness statements, operational statistics, logistics references, platform availability, weapons-use figures, and other facts that may interact. The purpose would not be to create a new classified archive of public information. It would be to preserve institutional memory about what the public record already permits an analyst to know.

Second, selected releases should receive an adversary-view reconstruction. A small OPSEC or intelligence red team could be given only publicly available U.S. material and asked a simple question: what could a competent foreign analyst estimate now that could not be estimated before this release? If the answer is “nothing material,” publication proceeds. If the new fact materially sharpens an estimate of capacity, timing, limitation, or dependency, the release can be delayed, generalized, or separated from another disclosure.

Third, the review should account for time. Operational sensitivity decays. A detail that would be risky during an active movement window may be harmless after the force has repositioned. A mosaic test should therefore favor delay and precision over permanent secrecy: protect the fact while it changes an adversary’s decision, then release it when that value expires.

Fourth, the test should be tiered. Most releases do not need an intelligence red team. The extra review should be reserved for information that touches operational tempo, force availability, munitions expenditure, logistics, deployment timing, maintenance, or dependencies. The goal is a better filter, not another clearance bureaucracy.

The advantage of a mosaic test is that it does not require the Pentagon to invent a new security theory. The theory is already embedded in DoD information-security and OPSEC doctrine. Classification guidance recognizes that compilation can change sensitivity. OPSEC doctrine recognizes that open indicators can be pieced together into estimates of critical information. The proposed change is procedural: apply that logic more systematically to the cumulative public record during fast-moving campaigns.

That approach also protects the value of public communication. A military that says nothing creates its own strategic costs. Allies need confidence. Citizens need evidence. Adversary propaganda should not be allowed to monopolize the first account of an operation. Commanders and public-affairs officers need room to explain what the joint force is accomplishing.

The objective is therefore not maximum secrecy. It is maximum awareness of the intelligence value created by aggregation.

The Pentagon already asks whether a fact is classified, releasable, and operationally sensitive. The next question should be what that fact becomes after it joins the facts released yesterday.

Public disclosure will remain part of modern warfare. So will adversary exploitation of public information. A mosaic test would help preserve the first while making the second harder.

Burak Oktenli is an independent researcher based in Washington, DC, focusing on authority, assurance, and governance in autonomous and AI-enabled systems. He holds a bachelor's degree in computer science and engineering from the University of South Florida and an MBA, and is completing a Master of Professional Studies in Applied Intelligence at Georgetown University. His analysis has appeared in RealClearDefense, the Washington Examiner, the Modern War Institute at West Point, RUSI, and Articles of War.