It started with a scroll, harmless enough in the quiet rhythm of an evening feed. Somewhere between the memes and the mundane, a video popped up featuring a man in military uniform, looking straight into the camera, speaking with the gravity of a whistleblower. He identified himself as a Polish soldier, and his words were incendiary. He talked about Ukrainian refugees with contempt, mentioned dark chapters of history, and warned of a wave of migrants that the government was hiding. It was the kind of content that makes a person stop scrolling, heart pounding, wondering if they were finally seeing the truth that the mainstream media wouldn’t touch. But here is the catch: the soldier wasn’t real. He never served in the Polish army. He never existed at all. He was a digital phantom, a fabrication born from artificial intelligence, designed specifically to look, sound, and behave like the real thing. And in recent days, these phantom soldiers have been flooding the Polish segment of social media, appearing on YouTube, TikTok, Facebook, and Instagram in what Ukraine’s Center for Countering Disinformation has flagged as a coordinated, large-scale operation to tear apart the fragile fabric of trust between Poland and Ukraine.
The videos themselves were carefully crafted to wound. Styled as interviews, they featured these AI-generated men railing against Ukrainians with a venom that felt disturbingly authentic. They dredged up the Volhynia massacres of World War II, a deeply painful historical wound between the two nations, weaponizing it in a modern context to stoke ethnic resentment. They repeated claims that Ukrainian refugees were driving up crime rates in Poland, tapping into anxieties already simmering in border towns and big cities alike. The fabricated soldiers used derogatory nicknames for Ukrainians, trying to normalize an ugly vocabulary of hatred. They leveled accusations against Poland’s Border Guard agency, suggesting corruption and incompetence, and predicted an impending invasion of migrants that would overwhelm the country. Amid all this, they wove in anti-European Union rhetoric, painting Brussels as an out-of-touch, oppressive bureaucracy that would sell Poland down the river. It was a unified, systematic attack designed to make Polish citizens distrust their own institutions, resent their neighbors, and question their place in Europe.
What gave the game away was a set of subtle, almost invisible fingerprints left behind by the people who made them. Poland’s NASK state research institute took the videos apart, analyzing them frame by frame and word by word, and found signs pointing to a possible Russian connection. The awkward phrasing in the Polish dialogue didn’t seem to come from a native speaker—it carried the unmistakable cadence of translation from Russian. Some of the videos had stray Cyrillic characters hiding in the metadata or in the background. And then there was the uniforms. The uniforms worn by these fake soldiers didn’t quite match Polish military attire; they more closely resembled the garb of Russian and Belarusian forces. It was as if the AI had been trained on thousands of images of the wrong army, and in doing so, accidentally exposed the hand behind the curtain. These small technical inconsistencies, invisible to the average viewer, became the critical evidence that turned a suspicious string of posts into a documented act of foreign interference.
The scale of the operation is staggering. Fact-checking outlet Konkret24 noted that while some accounts had already been taken down, one single YouTube channel had managed to post more than fifty of these fabricated clips before anyone could stop it. The most popular among them had nearly three hundred thousand views, meaning a substantial portion of the Polish public had likely been exposed to this poisonous content. Even conservative math suggests a reach that would make any political operative envious. And this isn’t an isolated incident—it’s a recurring pattern. Experts have watched this tactic evolve over time. In October, a video surfaced showing purported Polish soldiers calling for the seizure of Lviv, a Ukrainian city with deep historical ties to Poland. That video had been generated using a specific AI model called Veo, and its watermark was accidentally left visible, providing undeniable proof of its synthetic origin. A month later, another clip went viral, this one showing a crying twenty-three-year-old man who claimed to have been forcibly mobilized into Ukraine’s army. But again, inconsistencies in his uniform betrayed the deception—the details didn’t add up, and Polish state news agency PAP confirmed the clip had been artificially generated.
Taken together, these incidents paint a clear picture of a war happening in the information space that runs parallel to the physical war on the ground. The goal is not to change one person’s mind in a single moment, but to plant seeds of doubt, to corrode trust slowly, and to exploit the divisions that already exist. Poland and Ukraine have historically had a complicated relationship, but Russia’s full-scale invasion of Ukraine in 2022 brought them closer together than they have been in decades. Poland has served as a critical hub for Western military aid flowing into Ukraine, and has taken in millions of refugees. It is precisely this solidarity that the disinformation campaign aims to destroy. If Russia can make Poles believe that Ukrainians despise them, that the Ukrainian government is unstable and full of criminals, and that the European Union is manipulating them both, then it can erode public support for continued aid. One of the most heartbreaking aspects of this psychological warfare is that it treats real human beings—refugees fleeing bombs, soldiers defending their homeland—as nothing more than props in a cruel narrative game.
But there is a silver lining, however fragile. The fact that fact-checkers, state research institutes, and international partners are publicly identifying these fake videos is itself a victory. The disinformation machine depends on secrecy and confusion; when its works are exposed, it loses its power. The Center for Countering Disinformation, along with Polish fact-checking platforms like DemagogPL and Konkret24, are doing the difficult, unglamorous work of debunking lies in real time, hoping that a citizen who sees two conflicting videos will choose to trust the verified truth rather than the sensational fiction. Still, the fight is exhausting and ever-evolving. Every time a fake video is debunked, the AI models get better, the production quality improves, the tells get harder to spot. It’s an arms race in which the weapons are pixels and paragraphs, and the frontline is the human mind. For the ordinary person, the lesson is humbling and unsettling: what looks like a soldier’s confession may be a sophisticated lie, what seems like authentic journalism may be a weaponized narrative, and in this new era of synthetic media, the most essential skill may simply be the willingness to pause and question before believing.