The First 36 Hours of a Bioweapon Outbreak Are an Intelligence Crisis

Annie Jacobsen puts the central problem in blunt operational terms: experts across the agencies she studied told her that a biological outbreak might be stopped before becoming a pandemic if it were contained within 12 to 24 hours, perhaps as long as 36. But in the logic of her scenario, those same first hours are when U.S. officials have the least certain picture of whether a release occurred, what agent was involved, and whether the host government is concealing it.

Her hypothetical begins with an explosion at Vector, a Russian BSL-4 laboratory in Koltsovo, a science city about eight miles from Novosibirsk. Vector was a major site of the Soviet bioweapons program during the Cold War, she says, and it had an explosion in 2019, though no release was known to have resulted. Jacobsen changes that condition: a blast releases genetically modified pneumonic plague.

Novosibirsk is Russia’s third-largest city, Jacobsen says, with 1.6 million people. High-containment laboratories are often near major population centers, she adds, naming Tokyo, Atlanta, and locations in South Africa.

The crucial complication is political as much as biological. Biological weapons are illegal, Jacobsen says, so a state that possesses them—or loses control of them—has reason to deny both the program and the accident. A request through State Department counter-proliferation channels might therefore be met with silence rather than a usable warning.

That turns containment into a decision problem before it becomes a medical one. In Jacobsen’s scenario, officials must determine whether there was a release before they know the agent, its spread, or whether the state at the center of the event will cooperate. The clock that makes early intervention valuable also makes waiting for confirmation dangerous.

The precedent Jacobsen cites is the 1979 Sverdlovsk anthrax incident. Soviet personnel were milling anthrax into powder for weapons, she says, when an air-filter failure allowed it to escape. About 100 people in the town died. The Soviet explanation was contaminated meat, and the government denied a laboratory release for decades.

For Jacobsen, the episode establishes the likely pattern: concealment and misdescription at precisely the moment outside governments need prompt disclosure. It also distinguishes the threat from the pathogen in her hypothetical. Anthrax, she says, can be fatal but is a “dead-end host”: an infected person does not transmit it onward. Plague is a bacterium, not a virus, and pneumonic plague is the agent she selects for a release capable of becoming a wider crisis.

An explosion itself would likely not go unnoticed. Chris Williamson asks whether U.S. satellites would see it; Jacobsen says Defense Department officials told her they would have “eyes on” Vector. The site is a few miles from an ICBM facility near Novosibirsk, which she says gives officials another reason to monitor it. Satellite systems built to recognize missile launches could distinguish a laboratory explosion from an ICBM launch, she says, but measuring a blast from space is not the same as confirming that a pathogen escaped.

The proposed intervention may be more catastrophic than the suspected release. Williamson raises the prospect of destroying the site before the agent spreads. Jacobsen says that is plausibly a real, rather than merely cinematic, dilemma—but a nuclear strike on Russia would almost certainly trigger a nuclear exchange.

She calls this a Defense Department “wicked problem”: one in which a solution creates further problems, and new consequences have emerged by the time the solution is deployed.

The intelligence response Jacobsen describes has three questions: did a release occur, is there evidence of containment activity, and what is the pathogen?

The first question is whether people are dying and whether Russian government behavior indicates Moscow knows an accident took place. The National Reconnaissance Office, or NRO, supplies the “eyes” through space-based surveillance; the National Security Agency supplies the “ears” through signals intelligence, or SIGINT. Neither alone establishes that a biological agent escaped.

The second question is whether observable signatures point to an emergency response. Jacobsen highlights Measurement and Signature Intelligence, or MASINT, which looks for measurable changes. Sensors might be located in or on infrastructure and objects such as sewer pipes, walls, signs, drones, vehicles, or sprinkler systems. She frames their placement conditionally—“we have sensors, or maybe we have sensors”—but describes the purpose as detecting changes that could accompany efforts to mitigate an aerosol release.

Those changes could include industrial decontamination compounds. Jacobsen loosely compares them with much stronger versions of household cleaning agents such as ammonia or rubbing alcohol. Public spraying could be one of the clues intelligence officers watch for as they assess whether authorities are trying to contain something.

The third question is what escaped. Officials would need to determine whether they faced a virus or a bacterium, then identify the specific pathogen. In Jacobsen’s Vector scenario, plague is bacterial.

The system she describes is not a single alarm that produces a clean answer. It is an intelligence and defense apparatus trying to infer a hidden event from deaths, communications, imagery, and chemical signatures while the government with the best immediate information may be withholding it.