The Deepfake Case Files: the war that shipped with its own footage
In the first two weeks of the 2026 war, AI-fabricated combat videos drew tens of millions of views. This is the archive of the second war — the one fought inside your feed.
The 2026 Iran war had two fronts. The first one you saw on the news: missiles, air defenses, statements, maps with red arrows. The second front lived in your pocket, right where the thumb scrolls. And on that front, almost nothing was what it claimed to be.
CNN published the number on March 11: AI-generated combat videos racked up tens of millions of views within roughly the first two weeks of the war. No genuine field report came close. That is not a mystery; it is the physics of attention. Reality has to happen first. A fake only has to render.
This page is the hub of our deepfake case files. Its job is not to be outraged; its job is to be an archive — case by case, sourced and dated. In an information war, memory is the only air-defense system that actually works.
Case one: the flood
For two weeks, the feed rained explosions no camera had ever recorded: cities that were not burning but "burned", defense systems that failed or broke records depending on who ordered the clip. The point is not that audiences were gullible. The point is that the fabricated image was engineered for precisely the emotion the real news could only partially deliver.
Case two: the factory
The Foundation for Defense of Democracies reported on March 19 that much of this content came from networks linked to the Iranian government, amplified by Russian and Chinese information ecosystems. A state fighting an actual war still had budget for an imaginary one. Admire the prioritization: if victory is unavailable on the battlefield, it remains available in the render queue.
Case three: the counter-operation
The other side was not idle. The University of Toronto's Citizen Lab documented an Israeli-backed network — codename "PRISONBREAK" — pushing AI-generated anti-government propaganda at Iranian users. The full file is at The PRISONBREAK network. For now, one sentence suffices: both sides of the war were firing at the same screen. Yours.
Case four: verification, killed in action
What did the worried user do? Asked an AI whether the video was real. And Grok and Gemini, per Euronews on March 30, authenticated fakes as genuine. The fact-checking bot joined the casualty list. The details of that dark comedy are in When AI verifies AI.
Case five: the invoice
The flood was not free — it was profitable. Engagement pays out, and war is the highest-engagement product in the attention market. We unpack the business model in The war-slop economy.
Case six: the audience behind the wall
And the bitterest file: the primary audience of both fabrication factories was a population that had to climb over a censorship wall just to reach this poisoned internet in the first place. That story is VPN nation.
Analysis: you are not the audience, you are the terrain
The whole archive compresses into one sentence: in an information war, no side is your side. Tehran faked for you, Tel Aviv faked for you, Moscow and Beijing handled distribution, and the platform took a commission from everyone. The single common denominator of every operation was the target: the person holding the phone. Next time an image makes your blood boil, pause and run it through the seven checks in the war headline toolkit. Your nervous system is the battlefield. At least do not walk onto it unarmed.
The map of these files
This hub is alive and grows with every new case: The PRISONBREAK network, When AI verifies AI, The war-slop economy, and VPN nation. And for what respectable legacy media manage without any AI at all, read The victim hierarchy.
