Russian military uses AI to create fake videos of progress at the front

Author, Oleg Chernysh

Place of work, BBC Ukrainian Service, Kyiv

Published 2 minutes ago

Reading time: 6 min

Units of the Russian army have begun to actively use artificial intelligence (AI) tools to create fake photos and videos that are supposed to demonstrate their “advancement” at the front.

This is a translation of a text by a correspondent of the BBC Ukrainian Service. Read the original in Ukrainian here

Sources of the Ukrainian BBC service among the command of the Armed Forces of Ukraine confirm this trend. However, they note: at the moment, the main tactic of Russian troops remains the so-called “infiltration.” Separate groups of two or three, less often four or five, fighters try to penetrate between Ukrainian positions to the rear, secretly position themselves in a building or shelter, and after some time demonstrate the Russian flag on the UAV camera.

After this, the Russian Ministry of Defense puts this settlement on the list of “captured” ones, even if Ukrainian search and strike groups subsequently destroy “infiltrated” opponents.

However, when even small groups of attack aircraft cannot be brought to a populated area, the Russian military resorts to falsifications. Using AI, they “draw” their fighters on footage of a supposedly captured village or city.

“And then it turns out that the Russian military “took” those villages that are located at a distance of 7–10 km from the real front line. And we didn’t even see them there,” says a BBC interlocutor from the Sumy direction.

Similar fake messages about the “advancement” of Russian troops were also published from the Donetsk, Kharkov, Zaporozhye and Liman directions. As for the latter, a striking example is a video demonstrating the “presence” of a Russian soldier in the village of Svyatogorsk, the capture of which was even announced by Russian President Vladimir Putin on September 1.

Signs of AI are clearly visible in the footage - a soldier with a covered face is “embedded” in the landscape of the Svyatogorsk street.

The American think tank Institute for the Study of War (ISW) has researched this topic and identified in recent months dozens of examples of Russian military officers and bloggers publishing such videos about the “successes” of the Russian army.

What does the ISW report say?

ISW believes that such photos and videos are being taken to create the impression that Russia has the initiative at the front, and to make the situation appear as if the Ukrainian defense line is collapsing throughout the theater of operations.

“The creation of AI-edited videos appears to be an initiative of individual Russian military commanders and specific units, rather than a large-scale, centralized or standardized military practice,” the summary says.

ISW found that Russian sources began mass publishing AI-processed videos of flag raisings beginning in June 2026.

Between June 1, 2026 and August 30, 2026, ISW recorded at least 35 videos that were suspected of using AI. Commercially available software for identifying such content determined with a high degree of confidence that at least 10 videos were altered using artificial intelligence.

The total number of such videos is likely higher given the limitations of AI detection tools: they are not always successful at recognizing changes that are obvious to a trained observer.

ISW first recorded videos of Russian flags that had been altered by AI on June 11, 2026, and by the end of that month, Russian sources had released at least eight such videos that were likely created using artificial intelligence.

The Institute for the Study of War recorded that in July and August, Russian sources published at least 13 and 14 videos, respectively, containing footage altered by AI.

ISW believes the Russian military is using AI-processed material as part of a cognitive warfare campaign to portray Russian victory in Ukraine as inevitable.

Commercial software for detecting AI-generated content shows that the analyzed videos were created using Sora and Seedance models.

Sora and Seedance are generative AI models that create videos based on input text or reference material.

Sora was released in 2024 by American artificial intelligence company OpenAI.

Seedance was released a year later by the Chinese artificial intelligence and social media company ByteDance.

Examples of fake videos

On July 27, the Russian pro-war telegram channel “Divgen” published a video of a Russian soldier raising a flag on the eastern outskirts of the village of Belyi Kolodez in the Kharkov region.

Russian and Ukrainian sources reported that the video was edited using AI, taking into account the soldier's unnatural physical proportions and his sudden disappearance from the frame.

The Russian Ministry of Defense announced the capture of this settlement on August 3.

On August 24, Z-blogger Voin DV, who often publishes footage from the zone of the Russian group of troops “Vostok”, published a video in which Russian military personnel with the same disturbed body proportions raise flags in Dolinka in the Zaporozhye region.

On the same day, the Russian Ministry of Defense announced the capture of this village.

ISW notes that Russian sources use panning and zooming in their videos to reduce the amount of time that AI-generated or altered military personnel are visible in the frame, thereby making it more difficult to detect anomalies.

The footage, which was likely edited using this technology, contains splices, erratic zooming or panning, and inconsistencies in the shadows near depictions of Russian troops and flags.

Editing is used to link two scenes: a close-up of a Russian soldier waving a flag and a wide shot of the locality where the event supposedly takes place.

The anomalies revealed suggest that AI models are still having difficulty creating convincing images of Russian military personnel or animating complex scenes.

It is likely that Russian sources are filming videos in front-line settlements precisely in order to use them as source material for AI models. Some pro-Russian “military correspondents” added maps or landmarks to their AI-processed videos indicating the location of the flag raising to facilitate geolocation.

Detecting AI manipulation in such videos requires a more thorough analysis of the troops or flags depicted, rather than detailed examination of geolocation data, the Institute for the Study of War notes.

The Institute for the Study of War has identified at least 10 regiments and brigades that appear in videos that are assessed to have been or are suspected of being edited by AI.

In addition, such videos were used by the Russian command for reports.

On August 14, a Russian source published a video that allegedly showed Russian troops holding positions in several settlements liberated by Ukrainian troops.

AI-generated content detection software determined with a high degree of confidence that scenes allegedly depicting Russian soldiers in Novonikolayevka and Zlagoda (Dnipropetrovsk region), as well as in the northern part of Privolny, Rybny, Krasnogorsky and Novogrigorivka (Zaporozhye region), were altered using AI models.

Russian authorities have not commented on cases of their military using videos edited with the help of AI.