On Sept. 1, Russian President Vladimir Putin announced that Russian forces had captured Sviatohirsk, a town anchoring the northern flank of Ukraine’s Sloviansk-Kramatorsk fortress belt. Six days later, Brig. Gen. Andrii Biletskyi stood in the town and filmed himself speaking calmly. Even a Kremlin-affiliated military blogger conceded it proved Ukrainian control. A pro-war channel that had posted two “liberation” clips later admitted a neural network produced them.
The lie lasted less than a week. But Sviatohirsk was not an isolated blunder. It was one of at least 35 videos that the Institute for the Study of War suspects were AI-altered between June 1 and Aug. 30. Washington should read the pattern correctly, because it is not about fooling the public. It is about making the truth expensive.
Fabricated victories are as old as war. What artificial intelligence changes is the price. A flag-raising clip that once required a risky infiltration now takes a few minutes of generation. Ukrainian officials have described the earlier, deadly version. Spokesman Viktor Trehubov recounted how Russian propaganda claimed the village of Bilyi Kolodyaz after two infiltrators tried “the classic flag-plant tactic,” then fled or died. Days later, a milblogger posted a clip of a soldier waving a flag there. The ISW assessed it as AI-generated, noting the soldier vanished as the drone camera pulled back.
The Kremlin appears to be getting this almost by accident. The ISW says the practice doesn’t look centrally directed. Individual units and bloggers are producing it, possibly to hide failures or to show that Moscow’s objectives were met. That fits a system in which careers depend on reporting captures. Ukraine’s 19th Army Corps commander said in June that Russian commanders had already reported Kostiantynivka’s capture up the chain and were scrambling to make it true. Ukraine’s Center for Countering Disinformation put it bluntly: The videos let commanders “report” success to the Kremlin virtually.
Here is the problem for everyone else. Verification has become the battlefield’s referee. Satellite imagery, drone footage, and geolocation let open-source analysts check claims within hours. AI attacks that system not by erasing real evidence but by contaminating it. The ISW warns that Russian sources may film real settlements and use the footage as a reference for generated scenes. The street is real. The buildings are real. The soldiers are fiction. Geolocation, long the gold standard, no longer settles the question.
Commercial detection tools offer limited relief. They flagged at least 10 of the 35 videos with high confidence, and the ISW says the true number is likely higher because the software spots manipulation inconsistently. Some fakes are laughably crude, such as a flag apparently sprouting from a rooftop. Each one still demands a human analyst. Moscow can produce another claim faster than Kyiv, and its Western partners can disprove it.
That asymmetry is now being aimed at Ukrainian successes. When Ukrainian President Volodymyr Zelensky announced in August that Ukraine had liberated 745 square kilometers and 26 settlements, Russian sources responded with footage purporting to show Russian troops still holding several of them. The ISW assessed that some of it was AI-altered. A false capture can be exposed by events, as Sviatohirsk was. Counterfootage against a Ukrainian advance only needs to leave the record muddy. Biletskyi’s latest claim, that Operation Vivaldi cleared 85 square kilometers north of Lyman, is the obvious next target.
There is a consolation, and Americans should not overstate it. The same culture of lying is poisoning Russia’s own military. The ISW assesses that a “pervasive culture of lying” degrades Russian combat performance. Even Russian bloggers have noticed: forces cannot call in strikes on positions they claim to already hold, and the channel Rybar warned that deliberately false reports could lead to operations planned on a fantasy. A commander who believes his own paperwork is a commander who gets his men killed.
But Washington cannot count on Moscow’s self-deception to carry the day. The broader trend is visible in Anthropic’s September threat-intelligence report, which documents Russian state-linked actors using AI to speed up espionage and influence operations, including media workflows. The report doesn’t tie those operations to the flag videos, but it shows the same economics at work: tasks that once needed teams of specialists now run at machine speed.
GEORGIA’S OPPOSITION HATES TRUMP MORE THAN THEY FEAR PUTIN
What should the West do? Better deepfake detectors will help, but they are not the answer. The answer is verification capacity: more analysts fluent in the languages and platforms, faster access to commercial satellite imagery, provenance standards for footage, and trusted chains of evidence that allies can share quickly. Watch for three signs that the tactic is hardening into doctrine: Russia’s Defense Ministry publishing suspected synthetic footage itself, counterfootage aimed at Operation Vivaldi, and AI-generated satellite imagery.
The decisive question in this phase of the information war is not whether AI can fool an audience. It is whether democracies can authenticate reality faster than Moscow can manufacture doubt. Right now, the cost of lying is falling faster than the cost of checking. That is a gap worth closing before the next flag goes up.
Maksym Panchokha is a senior at Bard College studying Economics and Global & International Studies, with experience in research and policy analysis at the Foundation for Defense of Democracies, the Permanent Mission of Ukraine to the United Nations, and the Wilson Center’s Kennan Institute.
Continue reading...
[ H/T Washington Examiner ]
