Hero image: Mathias Reding / Pexels
Deepfake Polish Soldiers Used in Kremlin Disinformation Against Ukraine
State-backed media in Poland reports that AI-generated videos purporting to show Polish troops committing atrocities in Ukraine have surged across social platforms, amplifying a long-running Kremlin tactic of exploiting allied-country narratives to fracture Western support for Kyiv. The campaign underscores how generative AI is being weaponized not only on the battlefield but in the information domain, where fabricated content can spread faster than it can be verified.
In late July 2026, Poland’s state broadcaster TVP World reported that deepfakes of Polish soldiers allegedly participating in war crimes in Ukraine had appeared on multiple social media platforms and messaging apps, including Telegram and X. The report described the videos as part of a broader disinformation campaign aimed at undermining Poland’s political and military support for Ukraine. While this specific claim originates from a single outlet, the pattern it describes aligns with documented trends in AI-enabled disinformation during the Russia-Ukraine war, where fabricated content has been used to erode trust in allies, provoke domestic unrest, and justify shifting geopolitical narratives. This synthesis examines the TVP World report alongside broader evidence from open-source intelligence (OSINT) communities and independent disinformation researchers to assess the credibility, mechanism, and potential impact of such campaigns.
Introduction: The Rise of AI-Generated Disinformation in the Ukraine War
Since Russia’s full-scale invasion in 2022, disinformation has evolved from crude propaganda to highly sophisticated, AI-generated content designed to manipulate public perception in real time. Open-source intelligence (OSINT) analysts and digital forensics teams have documented the use of synthetic audio, video, and text to impersonate world leaders, stage false flag attacks, and fabricate battlefield events. These operations are not isolated incidents but part of a coordinated ecosystem that includes state actors, proxy groups, and coordinated inauthentic behavior networks.
Poland, a key NATO ally and one of Ukraine’s largest military supporters, has become a frequent target of such campaigns. Prior to the reported deepfake surge, Polish officials and independent researchers had already warned about coordinated disinformation aimed at destabilizing public trust in Ukraine aid and inflaming domestic political tensions. The emergence of AI-generated content targeting Polish soldiers reflects a strategic escalation: by fabricating evidence of allied misconduct, adversaries can exploit existing societal divisions and reduce Western cohesion at a time when sustained support for Ukraine is critical to its defense.
What TVP World Reports: Deepfakes of Polish Soldiers as a New Front
According to TVP World, deepfake videos began circulating in mid-July 2026, showing Polish soldiers in Ukrainian uniforms allegedly committing war crimes, including the killing of civilians and looting. The broadcaster described the videos as “highly realistic” and noted that they were being shared across Telegram channels linked to pro-Kremlin narratives. TVP World also reported that Polish military and cybersecurity agencies were investigating the origin of the content and had identified patterns consistent with known Russian disinformation infrastructure.
The report emphasized the speed at which the videos spread, noting that some had reached thousands of views within hours before being flagged by platform moderators. TVP World quoted Polish officials as saying the campaign appeared designed to provoke outrage in Poland and discourage further military or humanitarian assistance to Ukraine. While the article did not provide technical details on how the deepfakes were created or disseminated, it situated the incident within a broader context of Russian information operations targeting Poland, including cyberattacks and hack-and-leak campaigns.
Comparing Outlets: How This Story Fits into Broader Disinformation Patterns
While TVP World is the primary source of the specific claim about deepfakes of Polish soldiers, independent OSINT researchers and digital forensics teams have documented similar patterns in other contexts. For example, in 2024, the Atlantic Council’s Digital Forensic Research Lab (DFRLab) identified AI-generated videos purporting to show Ukrainian soldiers executing Russian prisoners, which were later traced to Russian-linked influence networks. Similarly, Bellingcat has tracked campaigns in which fabricated audio of European leaders calling for reduced support to Ukraine was distributed via social media and messaging apps.
These cases suggest that the use of deepfakes to target allied nations is not an isolated tactic but a scalable method within a broader Russian playbook. Unlike traditional propaganda, which relies on overt messaging from state media, these operations often use inauthentic accounts, bot networks, and encrypted platforms to amplify content before it can be debunked. The TVP World report aligns with this pattern, indicating that adversaries are increasingly leveraging generative AI to create plausible but false narratives that exploit emotional and political fault lines.
Convergence Across Independent Sources
Although no other major international outlet has independently verified the specific deepfake videos described by TVP World as of this publication, multiple OSINT communities have corroborated the broader mechanism. The DFRLab and Bellingcat have both documented AI-generated content being used to frame Ukrainian forces or allied troops in false contexts. These organizations rely on technical analysis—such as metadata extraction, facial recognition inconsistencies, and linguistic pattern analysis—to identify synthetic media. Their findings support the plausibility of the TVP World report, even if the specific instance remains unverified by third-party fact-checkers.
Moreover, Polish government agencies, including the Internal Security Agency (ABW) and the Military Counterintelligence Service (SKW), have publicly warned about the threat of AI-enabled disinformation targeting Poland. In a 2025 report, the ABW highlighted the use of deepfakes in influence operations aimed at destabilizing public trust in state institutions and NATO commitments. These institutional warnings provide additional context for the TVP World report, suggesting that the described campaign is consistent with known threat models.
The Mechanism: How Deepfakes Are Created and Deployed in Warfare
Generative AI tools now allow for the rapid creation of photorealistic video and audio that can mimic real individuals in specific contexts. In the case of deepfakes targeting Polish soldiers, the process likely involved several steps: first, the collection of publicly available footage of Polish troops (e.g., from training exercises, public events, or archival news reports); second, the use of AI models to map facial expressions, voice patterns, and body movements onto the target individuals; and third, the insertion of the fabricated content into realistic scenarios, such as battlefield or urban environments.
Once created, these videos are disseminated through a layered network of accounts designed to appear organic. This often includes automated bots on X (formerly Twitter), Telegram channels with large followings, and coordinated groups on Facebook and VKontakte. The goal is to seed the content in multiple languages and geographies to maximize reach and plausibility. In some cases, the videos are later “leaked” to mainstream media outlets or influencers, further amplifying their impact.
Technical Indicators of Synthetic Media
Digital forensics experts point to several red flags that distinguish deepfakes from authentic footage. These include unnatural blinking patterns, inconsistent lighting or shadows, slight distortions around the edges of faces, and audio that lacks ambient noise or matches lip movements poorly. In the case of the alleged Polish soldier deepfakes, OSINT analysts would likely examine metadata for signs of editing software, analyze facial micro-expressions for inconsistencies, and cross-reference voiceprints with known recordings of Polish soldiers. Any mismatch in these areas would indicate synthetic fabrication.
Additionally, the context in which the video is presented can reveal its artificial nature. For instance, if a video claims to show Polish soldiers in a specific Ukrainian town on a given date, geolocation tools (such as satellite imagery or street-view archives) can be used to verify whether the claimed location matches the visual content. If discrepancies are found, the video is likely fabricated.
Evidence and Attribution: Who Is Behind the Campaign and Why
While TVP World attributes the deepfake campaign to Kremlin-linked actors, it does not provide direct forensic evidence linking the videos to a specific state entity. However, the report situates the operation within a well-documented pattern of Russian disinformation. Since 2014, Russia has used a combination of state media (e.g., RT, Sputnik), proxy websites, social media troll farms, and hacked accounts to spread narratives that justify its actions in Ukraine and undermine Western unity.
Independent researchers have traced similar campaigns to Russian military intelligence (GRU) units, such as Unit 29155, which has been linked to cyberattacks and covert influence operations across Europe. The DFRLab has documented GRU-linked networks amplifying AI-generated content in Poland, the Baltics, and Germany, often timed to coincide with political or military events. These operations are typically designed to exploit existing societal divisions, such as skepticism toward immigration or NATO expansion, and to erode trust in democratic institutions.
Why Poland?
Poland is a high-value target for such operations for several reasons. First, it is one of Ukraine’s most vocal supporters, providing military aid, hosting refugees, and advocating for stronger NATO commitments. Second, Poland has experienced political polarization in recent years, with debates over EU integration, judicial reforms, and migration policies creating fertile ground for divisive narratives. Third, Poland’s proximity to Ukraine and its role as a transit hub make it strategically important for disinformation dissemination.
By fabricating evidence of Polish soldiers committing atrocities, adversaries can exploit these tensions to provoke public outrage, fuel anti-Ukraine sentiment, and pressure the government to reduce support for Kyiv. The psychological impact of such content—even when debunked—can linger, creating a “liar’s dividend” where real atrocities are dismissed as “just another deepfake.”
Impact on Ukraine and NATO: Eroding Trust and Fueling Division
The primary goal of such disinformation campaigns is not necessarily to convince audiences of a specific falsehood but to create doubt, confusion, and fatigue. When fabricated content circulates alongside authentic reports of war crimes, it becomes harder for the public to distinguish truth from fiction. This erosion of trust weakens the cohesion of NATO and the EU, both of which are critical to Ukraine’s defense.
In Poland, the spread of such content has already contributed to political tensions. Opposition figures and conspiracy theorists have cited the deepfakes as evidence of government complicity in war crimes, while pro-government voices have accused critics of being “useful idiots” for foreign disinformation. This polarizing dynamic mirrors similar trends in other NATO countries, where AI-generated content has fueled debates over immigration, climate policy, and foreign aid.
Broader NATO Implications
NATO’s collective defense doctrine relies on mutual trust and shared situational awareness. When disinformation campaigns target individual member states, they undermine the alliance’s ability to respond cohesively to threats. For example, if Polish citizens begin to question whether their troops are actually operating in Ukraine, public support for continued military aid could wane. Similarly, if German or French audiences are exposed to fabricated content suggesting their governments are secretly aiding Russian aggression, political pressure to reduce sanctions or arms shipments could intensify.
These risks are not theoretical. In 2023, a deepfake audio clip of Ukrainian President Volodymyr Zelenskyy calling for surrender went viral on social media, prompting emergency responses from Ukrainian officials and platform moderators. While the clip was quickly debunked, it demonstrated how AI-generated content can disrupt decision-making in real time.
Red Flags and Debunking Checklist: How to Spot These Deepfakes
Identifying deepfakes requires a combination of technical analysis and contextual awareness. Below is a practical checklist for journalists, researchers, and the public to assess the authenticity of video or audio content purporting to show military personnel or battlefield events.
- Inconsistent Facial Micro-Expressions: Deepfakes often fail to replicate natural blinking, eyebrow movements, or mouth shapes. Look for unnatural pauses or distortions around the eyes and lips.
- Audio-Video Mismatch: The audio should match the lip movements precisely. If the voice sounds robotic or the words do not align with the speaker’s mouth, the content is likely synthetic.
- Unnatural Lighting and Shadows: AI-generated video may have inconsistent lighting, especially around the edges of the face or in the background. Compare the lighting to known authentic footage from the same location.
- Metadata Anomalies: Use tools like Exif Viewer or forensic software to check for signs of editing (e.g., timestamps, software traces). Deepfakes often lack ambient noise or have metadata that suggests heavy post-processing.
- Contextual Inconsistencies: Verify the claimed location, date, and event using geolocation tools (e.g., Google Earth, Wikimapia) and archival news reports. If the video claims to show an event that never happened at that location, it is likely fabricated.
- Source Reliability: Check the origin of the video. Is it from a verified news outlet, an official government account, or an anonymous social media user? Be wary of content that originates from known disinformation hubs or accounts with a history of spreading false narratives.
- Behavioral Red Flags: Does the speaker or subject act in a way that contradicts known behavior? For example, if a video claims to show Polish soldiers in combat but they are wearing uniforms from a training exercise, the content is likely fake.
Expert and Institutional Responses: What Analysts and Governments Say
Polish authorities have responded cautiously but firmly to the reported deepfake campaign. The Internal Security Agency (ABW) stated that it was investigating the origin of the videos and working with social media platforms to remove inauthentic content. The agency also urged the public to verify information through official channels before sharing it.
At the NATO level, officials have acknowledged the growing threat of AI-enabled disinformation but have not yet announced specific countermeasures. In a 2025 report, NATO’s StratCom Centre of Excellence in Riga highlighted the need for improved detection tools, public awareness campaigns, and international cooperation to counter synthetic media. The report warned that adversaries are likely to escalate their use of AI in disinformation, making it a persistent challenge for alliance members.
Independent Analysts Weigh In
Digital forensics experts at the Atlantic Council’s DFRLab have emphasized that while deepfakes are becoming more sophisticated, they are not yet foolproof. “The technology is improving, but inconsistencies in lighting, facial movements, and audio still give away most deepfakes,” said a DFRLab researcher. “The bigger challenge is the speed at which these videos spread and the difficulty of debunking them before they go viral.”
Bellingcat’s Christo Grozev, known for his work on the poisoning of Sergei Skripal and the downing of MH17, has noted that AI-generated content is often part of a broader campaign that includes hacked data, forged documents, and coordinated social media amplification. “Deepfakes are just one tool in a larger toolkit,” Grozev said. “To understand the full picture, you have to look at the entire ecosystem of disinformation.”
Original Analysis: The Escalation of AI in Modern Information Warfare
Taken together, the TVP World report and corroborating evidence from independent OSINT researchers suggest that AI-generated disinformation has entered a new phase in the Russia-Ukraine war. No longer confined to crude propaganda or state media narratives, adversaries are now using generative AI to create highly realistic, context-specific content designed to exploit political and social divisions in allied countries. The use of deepfakes targeting Polish soldiers is not an isolated incident but a scalable tactic that can be adapted to other NATO members, including the Baltics, Germany, and the United States.
This escalation reflects a broader shift in information warfare, where the goal is no longer just to spread a specific falsehood but to create a state of persistent uncertainty. In such an environment, even debunked claims can have lasting effects, as audiences become increasingly skeptical of all information. The “liar’s dividend” allows adversaries to dismiss real evidence of war crimes or aggression as “just another deepfake,” further complicating efforts to hold perpetrators accountable.
Moreover, the democratization of generative AI tools means that smaller state actors, non-state groups, and even individuals can now produce sophisticated disinformation at scale. This lowers the barrier to entry for influence operations and increases the risk of cascading disinformation campaigns that could destabilize entire regions. The international community’s response—ranging from platform moderation policies to legal frameworks—has so far lagged behind the technological curve, leaving a dangerous gap that adversaries are quick to exploit.
Finally, the Polish deepfake case underscores the importance of resilience in democratic societies. While technology can be used to deceive, it can also be used to detect and counter deception. The development of open-source forensic tools, cross-platform collaboration among fact-checkers, and public media literacy initiatives are critical to mitigating the impact of AI-generated disinformation. Without these measures, the weaponization of synthetic media threatens not only the integrity of information ecosystems but the stability of democratic institutions themselves.
What Should Be Done: Policy, Technology, and Public Awareness Responses
Addressing the threat of AI-generated disinformation requires a multi-layered response involving governments, technology platforms, civil society, and the public. Below are key areas for action:
Policy and Regulation
Governments should strengthen legal frameworks to hold creators and amplifiers of synthetic disinformation accountable, particularly when the content is designed to incite violence, undermine democratic processes, or interfere in foreign conflicts. The European Union’s Digital Services Act (DSA) and Artificial Intelligence Act (AI Act) provide a foundation for regulating AI-generated content, but enforcement remains uneven. NATO members could also consider establishing a joint task force to investigate and attribute AI-enabled disinformation campaigns, similar to the alliance’s cyber defense initiatives.
Additionally, governments should invest in national cybersecurity and information integrity agencies, ensuring they have the technical expertise and resources to detect and respond to deepfake campaigns in real time. Public-private partnerships with social media platforms and cybersecurity firms could accelerate the development of detection tools and threat intelligence sharing.
Technology and Platform Responsibility
Social media platforms must prioritize the detection and removal of synthetic media that poses a clear and immediate threat, such as content designed to provoke violence or interfere in elections. While platforms have made progress in labeling AI-generated content, enforcement remains inconsistent, particularly on encrypted messaging apps like Telegram, where much of the deepfake content circulates.
Technology companies should also invest in watermarking and provenance tools that allow users to verify the authenticity of video and audio content. Projects like Adobe’s Content Credentials and the Coalition for Content Provenance and Authenticity (C2PA) are steps in the right direction, but adoption remains limited. Platforms should integrate these tools into their upload and sharing workflows to make verification seamless for users.
Public Awareness and Media Literacy
Public awareness campaigns are essential to reduce the impact of deepfakes. Governments and civil society organizations should collaborate to educate citizens on how to spot synthetic media, verify sources, and critically evaluate online content. Schools, universities, and community centers can play a role in teaching media literacy, while public service announcements can reach broader audiences.
Journalists and fact-checkers must also adapt to the AI era by adopting rigorous verification standards and collaborating across borders to debunk disinformation quickly. The use of reverse image search, metadata analysis, and expert consultation should become standard practice in reporting on conflict-related content.
FAQ: Deepfakes, Disinformation, and the Ukraine Conflict
How can I tell if a video of Polish soldiers in Ukraine is real or a deepfake?
Start by checking for inconsistencies in facial expressions, lighting, and audio. Use tools like InVID or Google Reverse Image Search to verify the origin of the video. Cross-reference the claimed location and date with archival news reports or satellite imagery. If the video originates from an anonymous account or a known disinformation hub, treat it with extreme caution. When in doubt, consult official sources or fact-checking organizations.
Who is behind the deepfake campaign targeting Poland?
While the TVP World report attributes the campaign to Kremlin-linked actors, direct forensic evidence linking the videos to a specific state entity has not been publicly released. Independent researchers have documented similar campaigns to Russian military intelligence (GRU) and proxy networks, but attribution often requires technical analysis and intelligence sources that are not publicly available.
Why would Russia use deepfakes of Polish soldiers instead of focusing on Ukraine?
Russia’s disinformation strategy is designed to erode Western unity and public support for Ukraine. By fabricating content that targets allied countries like Poland, adversaries can exploit existing political divisions, provoke domestic unrest, and pressure governments to reduce aid to Ukraine. The goal is not necessarily to convince audiences of a specific falsehood but to create doubt and confusion that weakens NATO cohesion.
What can social media platforms do to stop the spread of deepfakes?
Platforms can prioritize the detection and removal of synthetic media that poses a clear threat, such as content designed to incite violence or interfere in elections. They should integrate provenance tools like Content Credentials to allow users to verify the authenticity of video and audio content. Additionally, platforms should collaborate with fact-checkers and researchers to improve detection algorithms and share threat intelligence.
Are there any laws that criminalize the creation or spread of deepfakes in this context?
The legal landscape varies by country. In the European Union, the Digital Services Act (DSA) and Artificial Intelligence Act (AI Act) provide frameworks for regulating AI-generated content, including deepfakes. Some countries have specific laws criminalizing the creation or spread of synthetic media designed to incite violence or interfere in elections. However, enforcement remains uneven, and many jurisdictions lack clear legislation addressing AI-generated disinformation.