Exposing a coordinated broadcast disinformation campaign

Background

Influence operations succeed on speed and deniability. A narrative can be seeded on a handful of outlets and echoed across many more in the time it takes a human analyst to finish watching a single segment. To the audience — and to the analyst — it looks like a spontaneous wave of opinion, which is exactly the point.

Most services still monitor broadcast media the way they did decades ago: analysts assigned to individual channels, watching by eye. That model can tell you what a channel said. It cannot tell you that nine channels said versions of the same thing, in the same window, traceable to the same origin.

The challenge

Coverage of a routine policy announcement turned sharply negative overnight. The same pointed framing — the same phrases, the same accusations, the same talking points — began appearing across multiple broadcast and web channels within the same evening.

The communications cell faced a question it could not answer by watching television: was this a genuine wave of public reaction, or a coordinated, seeded push designed to look like one? The distinction mattered enormously for how, and whether, to respond — over-reacting to organic criticism is its own kind of failure.

Monitoring the traditional way made the pattern almost impossible to see. Each analyst saw one hostile segment among many; no one could see that the segments were versions of a single, propagating narrative moving from a small set of origins outward.

The approach

• Pointed CLERINT Media at nine live TV and web streams simultaneously — transcribing speech to timestamped text, extracting on-screen tickers and chyrons through OCR, and tracking every face against a watchlist.
• Let automatic story clustering group segments that shared keyword patterns across two or more streams into a single emerging story, revealing the narrative as one object rather than nine fragments.
• Traced the propagation timeline to see where the framing originated and how it spread, distinguishing the two source streams that led from the seven that echoed them within the following ninety minutes.
• Built the narrative map, linking each recurring commentator to the stories they appeared in and the people they appeared alongside, exposing the connective tissue that tied the supposedly independent channels together.
• Set sentiment and face-driven alerts so the cell was notified in real time the moment the narrative shifted tone or a flagged individual reappeared on air.

For the first time we could see the campaign as a single thing, forming in real time, instead of arguing after the fact about whether it was coordinated at all.— Analyst, government communications cell

The outcome

Within the hour, the cell had evidence — frame-accurate and retained for the record — that the surge was not organic. The framing traced back to two originating channels and propagated outward on a timeline far too tight to be coincidence.

The narrative map showed a small set of commentators appearing across the supposedly independent channels, tying the coverage together into a single coordinated effort with identifiable connective tissue.

Armed with a defensible picture rather than a hunch, the cell could brief leadership on the nature of the campaign and respond to the coordination itself — not to each broadcast in isolation, and not by over-reacting to what would have looked like a groundswell of opinion.

Results

• Coordination established within the hour, with retained evidence for each segment.
• Two originating channels identified out of nine carrying the narrative.
• 27 recurring faces resolved into a single narrative map.
• Real-time alerting kept the cell ahead of the campaign as it evolved.

Why it matters

Coordination is invisible one screen at a time and obvious once every screen shares a graph. The capability that mattered was watching the whole surface at once and letting the shared structure surface itself.

Because the evidence was frame-accurate and retained, the finding was not just an internal judgement — it was something the cell could stand behind, show, and act on with confidence.