Russian sources are increasingly relying on AI-generated and AI-altered videos to assert territorial control over areas of Ukraine that Russian forces have not, in fact, seized. Russian sources are intentionally exaggerating territorial gains to portray Russian forces as advancing and having the initiative along the frontline and to falsely portray the Ukrainian line as collapsing across the theater.[1] The generation of AI-altered footage appears to be a grassroots effort by several Russian milbloggers and a subset of Russian units rather than a widespread, centralized, or standardized military practice. Certain Russian units may be publishing AI-altered material in order to claim to the Russian military leadership that they have met Kremlin-issued deadlines, or they may be using AI-altered footage to try to make exaggerated claims aimed at public audiences more credible. AI-manipulated footage is very likely here to stay, and Russian sources will continue to evolve their methodology to overcome commercially available AI detection software.

Russian sources began posting flag raising videos altered with artificial intelligence at scale in June 2026 to assert territorial control over areas of Ukraine that Russian forces had not seized.

Russian officials, units, and milbloggers have repeatedly amplified footage purportedly showing Russian forces holding flags and presented such footage as evidence of Russian forces’ seizure of settlements. Russian soldiers had been attempting to exploit OSINT mapping methodologies by having soldiers actually infiltrate areas and ostentatiously raise flags since at least September 2025.[2] The Russian sources are now evolving this information tactic by leveraging AI-altered footage to insert Russian soldiers with flags into geolocated videos to make such claims in places into which soldiers have not infiltrated. Russian sources combine authentic and AI-altered flag raising footage in video montages in an effort to pass off AI-generated videos as authentic.[3] ISW observed at least 35 videos suspected of AI alteration from June 1, 2026, to August 30, 2026. Commercially available AI detection software assessed with high confidence that at least 10 videos were altered with AI models. The total number of AI-altered videos is likely higher given the limitations of available AI detection tools, which do not always successfully identify AI alterations that are clearly visible to the trained observer. ISW first observed AI-altered flag waving videos on June 11, 2026, and Russian sources posted at least eight suspected AI-altered videos in June 2026.[4] ISW observed that Russian sources posted no fewer than 13 and 14 videos with AI-altered footage in July and August 2026 respectively.

Visual anomalies persist in suspected AI-altered videos despite efforts by Russian milbloggers to portray the footage as authentic.

Russian milblogger Divgen posted footage on July 27 showing a Russian servicemember raising a flag on the eastern outskirts of Bilyi Kolodyaz, Kharkiv Oblast. Russian and Ukrainian sources reported that the video was AI-altered given the improbable physical proportions of the Russian servicemember and their sudden disappearance from the scene.[5] A Russian milblogger who frequently publishes footage from the area of responsibility (AoR) of the Russian Eastern Grouping of Forces, Voin DV, posted footage on August 24 showing similarly misproportioned Russian servicemembers raising flags in Dolynka, Zaporizhia Oblast.[6] Russian milblogger Southern Republic posted footage on August 20 that showed a Russian flag disappearing or melting into a roof in Mykolaipillya, Donetsk Oblast.[7]

Russian sources use camera panning and zooming in their videos to reduce the length of time in which AI-generated or altered Russian servicemembers are visible to make it harder to detect anomalies. Suspected AI-altered videos contain jump-cuts, sporadic zooming or panning of the camera, and inconsistent shadows near the observed Russian servicemembers and flags.[8] Video cuts are used to make two scenes — a close-up shot of a Russian soldier waving a flag and an overview scene of the claimed settlement — appear connected. The observed anomalies indicate that AI models continue to struggle in generating persuasive images of Russian servicemembers or animating complex scenes.

Georeferencing footage manipulated in this way is not a viable practice for determining the authenticity of the scene. Geolocations derived from both suspected and assessed AI-altered videos are accurate because footage often actually does show parts of the claimed settlement.[9] Russian sources likely record videos of frontline settlements specifically to use as reference material for AI models. Several Russian milbloggers have included maps or reference points with their AI-altered videos to show the location of the flag raising events to support georeferencing.[10] Detection of AI manipulation in such videos requires closer scrutiny of the servicemembers or flags depicted, rather than closer interrogation of the geolocations.

Specific Russian military units appear to be producing or amplifying AI-altered footage to compensate for poor performance and their failures to secure military objectives on time. The use of AI-altered footage does not appear to be a widespread phenomenon across the Russian military for now. ISW has observed at least 10 regiments and brigades featured in assessed and suspected AI footage.[11]

A Russian milblogger repeatedly posted AI-altered footage of the 36^th Motorized Rifle Brigade (29^th CAA), likely to compensate for the brigade’s inability to make actual advances. Russian milblogger Voin DV has mentioned the 36^th Motorized Rifle Brigade in multiple assessed and suspected AI-altered videos since June 1, 2026.[12] Voin DV credited the 36^th Motorized Rifle Brigade with seizing seven settlements in Dnipropetrovsk and Zaporizhia oblasts over 10 kilometers ahead of the Russian forward line of own troops (FLOT). The 36^th Motorized Rifle Brigade has been operating in the Oleksandrivka Direction since November 2025.[13] Ukrainian military observer Kostyantyn Mashovets reported on March 4, 2026, that the command of the Russian 29th CAA (Eastern Military District [EMD]) investigated the 36th Motorized Rifle Brigade and found that the unit had exaggerated previous control-of-terrain (CoT) reports in the Oleksandrivka direction.[14] The 36^th Motorized Rifle Brigade likely relied on exaggerated control-of-terrain (CoT) reports to offset high casualty rates and territorial losses to Ukrainian counterattacks since January 2026.[15] Voin DV’s repeated attribution of battlefield successes to the 36th Motorized Rifle Brigade is likely furthering the brigade’s own efforts to conceal its failures.

Russian commanders have punished subordinates for leveraging AI-altered footage to offset poor performance. Russian milblogger Divgen credited the claimed seizure of Bilyi Kolodyaz to the 127^th Motorized Rifle Regiment (71^st Motorized Rifle Division, 14^th Army Corps [AC]).[16] A source reporting on the Northern Grouping of Forces claimed on July 30, 2026, that Russian Northern Grouping of Forces Commander Colonel General Yevgeny Nikiforov dismissed the commander of the 127^th Motorized Rifle Regiment and the commander of the regiment’s 1st battalion following the unit’s publication of AI-altered footage in Bilyi Kolodyazi, Kharkiv Oblast.[17] Russian milblogger Anatoliy Radov suggested that the original video was taken from a private chat used by the unit’s commander prior to modification.[18] The reported dismissal of the regiment’s commanders highlights the pressure placed upon Russian forces to make territorial gains. The Ukrainian 16^th Army Corps reported on January 30 that Nikiforov appointed a commission to evaluate the lack of progress in the Vovchansk direction.[19] The Ukrainian 16^th Army Corps reported on April 25^th that the 127^th Motorized Rifle Regiment conducted an unsuccessful motorized assault toward Vovchanski Khutory, Kharkiv Oblast.[20] Russian sources claimed on May 29 that the 127th Motorized Rifle Regiment seized Karaichne, Kharkiv Oblast.[21] ISW does not assess Russian forces to control Karaichne as of August 31, 2026.

Russian sources are publishing AI-generated and AI-altered footage to support claims about Russian advances in the Fortress Belt in Donetsk Oblast — the Kremlin’s priority operational objective. The Russian Ministry of Defense, Kremlin newswire TASS, official unit Telegram channels, and milbloggers mentioned eight Russian regiments and brigades operating in the Fortress Belt in suspected and assessed AI-altered footage. The Russian 3rd Combined Arms Army [CAA], formerly 2nd Luhansk People’s Republic Army Corps [LNR AC], Southern Military District [SMD]) and subordinate elements of the 123^rd Separate Motorized Rifle Brigade, 88^th Separate Motorized Rifle Brigade, 7^th Separate Motorized Rifle Brigade, and 1465th Motorized Rifle Regiment (4th Separate Motorized Rifle Brigade) were mentioned in four assessed and suspected AI-altered videos since June 1, 2026.[22] The official Telegram channel of the 3rd Combined Arms Army posted at least two suspected AI-altered videos. Commercially available AI detection software assessed with high confidence that at least two videos mentioning the 3rd CAA and subordinate elements were altered with AI models.[23] The 3^rd CAA’s combat effectiveness has degraded since June, and the formation can only conduct small-group infiltration efforts in the Slovyansk and Kramatorsk directions at this time.[24] Ukrainian 19th Army Corps Commander Brigadier General Alexander Bakulin reported on June 15^th that elements of the 3^rd CAA had personnel shortages and fear their commanders.[25] The Russian Ministry of Defense, the official Telegram channel of the 1442nd Motorized Rifle Regiment, and Russian milblogger Southern Republic mentioned the Russian 3rd Army Corps [AC] (Central Military District) and subordinate elements of the 1442^nd Motorized Rifle Regiment (6th Motorized Rifle Division) and 54^th Motorized Rifle Regiment in three assessed and suspected AI-altered videos. [26] Russian forces’ advances in the Fortress Belt have been slow and costly, and Russian forces are likely using this AI footage in order to advance a narrative that Russian forces are advancing more rapidly than they really are.[27]

Russian sources use AI-altered footage as part of a deliberate cognitive warfare campaign to portray a Russian victory in Ukraine as inevitable.

Russian sources claimed territory ahead of the actual Russian forward line of their own troops to give the impression of territorial advances where they do not exist. Russian Chief of the General Staff Army General Valery Gerasimov claimed on August 27 that Russian forces made gains in the Slovyansk-Kramatorsk direction and seized Pyskunivka, Donetsk Oblast (east of Slovyansk).[28] Commercially available AI detection software assessed with high confidence that several scenes purportedly showing Russian servicemembers in Pyskunivka were altered with AI models.[29] Gerasimov and other military commanders have exaggerated Russian gains in multiple public addresses since January 2026.[30] Russian President Vladimir Putin used a staged meeting with several Russian commanders on July 3 to make greatly exaggerated claims of Russian advances in Kostyantynivka, Donetsk Oblast, that do not match battlefield realities.[31] Russian sources posted at least four suspected or assessed AI-altered videos on July 3 and 4 purportedly showing Russian servicemembers waving flags in Kostyantynivka.[32] ISW continues to assess that the Kremlin likely uses these meetings as part of a cognitive warfare campaign to present Russia’s offensive as achieving its operational and strategic objectives.[33] Kremlin used the informational effect of the claimed seizure of Kostyantynivka to shape the United States administration’s perceptions and the Western information space. Russian Presidential Aide Yuriy Ushakov claimed on July 4 that Putin used a call with US President Donald Trump that day to emphasize the seizure of Kostyantynivka and claimed that Ukraine’s European partners have a false perception of the state of the frontline.[34] Putin remains committed to his theory of victory, which rests on the premise that Russia can win a war of attrition by having Russian forces maintain the initiative and advance across the entire theater indefinitely, thereby outlasting both Ukraine’s ability to defend and Western support for Ukraine.[35] The use of AI-altered footage to show fictitious gains supports this theory of victory and advances Russian efforts to persuade the US and the West to accept Russian claims of territory Russian forces have not seized. Most of the settlements claimed by Russian sources using assessed or suspected AI-altered footage remain under Ukrainian control as of August 31, 2026.

Russian sources used what is assessed and suspected to be AI-altered footage to deny demonstrated Ukrainian territorial gains in the Oleksandrivka direction. President Volodymyr Zelensky reported on August 12 that Ukrainian forces had concluded counteroffensive operations in the Oleksandrivka direction and liberated 26 settlements.[36] ISW assesses that Ukrainian forces liberated roughly 626.69 square kilometers in the Oleksandrivka direction and the northern part of the Hulyaipole direction (where Ukrainian forces also counterattacked) between January 29 and August 12, 2026.[37] Successful Ukrainian counteroffensives undermine Russia’s cognitive warfare campaign by demonstrating that Russian forces are unable to maintain the initiative and hold all territory they take. A Russian source posted suspected footage posted on August 14 purportedly showing Russian servicemembers maintaining positions in several settlements liberated by Ukrainian forces. [38] Commercially available AI detection software assessed with high confidence that scenes purportedly showing Russian soldiers in Novomykolaivka and Zlahoda, Dnipropetrovsk Oblast, as well as northern Pryvilne, Rybne, Krasnohirske, and Novohryhorivka, Zaporizhia Oblast, were altered with AI models. Geolocations of the suspected footage purportedly showed Russian servicemembers in five other settlements liberated by Ukrainian forces.[39]

Russian sources are leveraging publicly available video generation models to produce AI-altered videos.

Russian sources will likely continue to scale up their use of AI-altered footage given the ready accessibility of text-to-video generative AI models. Commercially available AI detection software indicates that assessed AI-altered footage was derived from Sora and Seedance AI models. Sora and Seedance are text-to-video generative AI models that develop videos from input text or reference material. Sora was published by American artificial intelligence company OpenAI in 2024.[40] Seedance was published by Chinese artificial intelligence and social media company ByteDance in 2025.[41] OpenAI discontinued the Sora web and app experiences on April 26, 2026, and reportedly intends to discontinue the application programming interface (API) on September 24, 2026.[42] OpenAI’s decision to discontinue Sora will force Russian sources to produce AI-altered footage with Seedance, or other AI models, including proprietary models the Russian government may develop for its own use.

Conclusion: Russian sources likely will continue to mature their use of AI-augmented footage to support cognitive warfare campaigns. ISW continues to assess that Russian sources are engaging in a sophisticated cognitive warfare effort that uses — among other means — AI-altered or generated footage of flag raisings to substantiate fabricated claims of Russian tactical successes in areas where Russian forces do not maintain positions.[43] This footage is designed to have geolocatable elements in order to imbue the videos with a veneer of legitimacy. Russian sources will continue to update their AI narratives and models to pass off fabricated gains as authentic. Text-to-video AI models provide Russian sources an inexpensive avenue to exaggerate Russian gains or deny the success of Ukrainian counteroffensives. Suspected AI-altered footage is increasingly able to bypass pixel content verification in commercially available AI detection software. Visual anomalies remain pervasive within AI-altered footage, however.

Russian sources may expand their cognitive warfare campaign to include AI-altered or generated satellite imagery. Google Earth’s Nano Banana AI tool allows users to generate AI-altered satellite imagery by text prompt.[44] Posts purportedly showing Nano Banana AI-altered or generated imagery indicate that users were able to bypass Google Earth’s terms of use by generating deepfakes of military sites. Russian sources will likely exploit similar AI models with limited guardrails to further their cognitive warfare campaigns. Open source researchers will need to develop and continuously refine responsible tradecraft to distinguish AI-altered footage or imagery from legitimate footage.