AI-generated videos showing young and attractive women promote Poland's EU exit

2025-12-3110:286147www.euronews.com

They are beautiful, eloquent — and do not exist. AI-generated girls from the 'Prawilne_Polki' profile called for Polexit and preached right-wing views. The TikTok account has been deleted, but…

AI-generated videos promoting Poland's exit from the European Union have appeared on Polish-language social media, featuring non-existent, attractive young women advocating for "Polexit".

One TikTok account called "Prawilne Polki" published content showing women dressed in T-shirts bearing Polish flags and patriotic symbols, European analytics collective Res Futura said. The content targeted audiences aged 15 to 25.

The videos featured statements including: "I want Polexit because I want freedom of choice, even if it will be more expensive. I don't remember Poland before the European Union, but I feel it was more Polish then."

Another video stated: "When I talk about Polexit, I hear: 'scaremongering', 'catastrophe', 'end of the world'. Always the same pattern. Zero conversation about who really decides for us and why. Maybe it's high time we started talking about it calmly."

Some videos appeared authentic while others showed clear signs of artificial intelligence use, with desynchronised vision and audio.

"The recordings were actually generated using a generative artificial intelligence tool. They are not of good quality at all," Aleksandra Wójtowicz, senior analyst for new technologies and digitalisation at the Polish Institute of International Affairs, told Euronews.

"If you look at them for a while, you can see that the facial expressions are unclear, the phrases spoken by the characters in the recordings are not completely natural."

The "Prawilne_Polki" profile has been removed, but similar accounts may emerge, Wójtowicz said.

"For now, it is not clear who is behind the content. At the moment there are several groupings that run various campaigns with the help of artificial intelligence," said Wójtowicz, who has researched online disinformation for years.

"They often take place on YouTube, but also on TikTok."

'Hydra effect' in motion

Similar content on YouTube imitates news services using synthetic voices or voices combined with generated faces, repeating pro-Russian narratives including claims that elections were rigged or Ukrainians are stealing, Wójtowicz said.

"From the research interviews I conducted, it appears that this AI mannerism, which for a large part is obvious, annoying, evident, for a certain audience is completely unnoticeable," Wójtowicz said.

The account was blocked after reports from organisations and individual users. Wójtowicz said TikTok may now better detect disinformation content about Poland potentially leaving the EU. When searching "Polexit", the platform displays a warning about disinformation.

"On TikTok it is quite typical that if such an action appears, there is a 'Hydra effect'. When these profiles are reported, several new ones may appear," Wójtowicz said.

"Now their creators know very well that this is inefficient, that it is not worth putting them up again when there is full interest in the phenomenon, because they will be deleted."

Wójtowicz outlined two possible scenarios: either the operation tested reactions, or disinformation actors began creating accounts to launch a campaign that was quickly detected. "Therefore, they will go underground and come back in the future."

Why young and beautiful women?

The content targeted young women specifically, Wójtowicz said.

"We have such a trend when it comes to the far right that young girls are targeted. And who better to appeal to young women if not other young women," she added.

"We are seeing phenomena on social media such as the tradwives movement, or traditional wives. There's a lot of this in the US, but it's making its way into Poland."

During Poland's presidential campaign, numerous AI-generated videos persuaded women to vote for right-wing candidates Sławomir Mentzen of Confederation or Karol Nawrocki, supported by Law and Justice or PiS party.

"This group is clearly very important, it can be a fundamental vector of political change. It is no coincidence that it was women who played such an important role in 2023," Wójtowicz said, referring to parliamentary elections in which PiS lost power after eight years.

'You can always count on the Russians'

"Righteous Poles! Russians can always be counted on. They forgot to translate 'prawilne'," Civic Platform MP, Jarosław Urbaniak wrote, referencing the Russian word "prawilnyj" meaning "proper" or "correct".

The term is used in Polish prison slang to describe a respected prisoner or someone who follows community rules.

Independent web developer Radek Karbowski said the content demonstrated the need for AI regulation. He calculated that the "Prawilne_Polki" profile generated 200,000 impressions and nearly 20,000 likes within two weeks, a 10% ratio that affects content spread and drives algorithmic recommendations.

The TikTok channel was established in May 2023 and operated under a different name, likely belonging to an English-speaking user publishing entertainment content unrelated to Poland, according to Polish portal Kontakt24.pl.

On 13 December, the account underwent a major transformation, receiving the new name "Prawilne_Polki" with a description reading: "Here speak beautiful Polish girls who speak straightforwardly, their own opinion. Patriotism, sovereignty and normality in one place. NO TO EUROCALCHOZ #polexit."

The takeover occurred when the account already had an established follower base. Three materials in the new style were published on 13 December: two videos with generated women and a graphic asking: "If the election were tomorrow - which party is number one for the right?"

Research conducted by UCE Research for Onet found that, in 2025, TikTok was the primary source of information about Poland's presidential election for 43.7% of respondents aged 18 to 25.

A United Surveys poll by IBRiS for Wirtualna Polska dated 22 December showed 24.7% of respondents favoured Polexit, while 65.7% opposed it.

A survey from early December conducted by French magazine Le Grand Continent showed 25% of Poles supported Polexit, while 69% opposed such a move.


Read the original article

Comments

  • By TrackerFF 2025-12-3114:421 reply

    Someone (Russia) is working HARD to spread anti-EU propaganda in Poland. It's a barrage of GenAI, bots, trolls, and what not.

    At least to me, living in Europe, the writing on the wall is that anonymous social media - or more likely internet as a whole - is going to end over here.

    I figure this is why some platforms, like X, are trying so hard to leverage their political power. They know that some verification / KYC measures are coming, and that they'll get fined or blocked if they can't follow those regulations.

    • By jech 2025-12-3114:591 reply

      > Someone (Russia)

      One of the videos uses the non-existent word prawilny, which is Russian (правильный). The Polish equivalent would be prawidłowy or właściwy.

      • By yetihehe 2025-12-3116:101 reply

        You are right, but it should be added that "prawilny" is used as a slang for "good guy" among some young people subcultures. It's no longer used much.

        • By jech 2026-01-0116:41

          There's certainly a number of Russian words in older Warsaw slang (barachło, ustrojstwo, wierchuszka, etc.), but the videos were not using slang, especially not older slang, and they had no reason to use prawilny except by accident.

  • By avaer 2025-12-3115:101 reply

    I wish governments would treat things like this as equivalent to fraud.

    It might not be taking people's money, but it's taking away access to truth through automated deception, which might even be worse.

    Back when it took a madman to pollute discourse the public square, maybe you could have called it their right free speech, but this is not that. This is just computer assisted fraud, and exactly the kind of thing that computer hacking laws should be for.

    • By eudamoniac 2025-12-3116:062 reply

      How is this different from hiring actors to do guerilla marketing, or hiring influencers to pretend to like something, both of which happen constantly?

      • By rchaud 2025-12-3116:29

        Time and money. There are laws limiting email spam because the unit cost of sending and distributing them is almost zero, which is a great incentive for bad people. Hiring actors and shooting videos for political propaganda takes time and money, whereas AI slop is nearly free.

      • By armchairhacker 2025-12-3116:27

        Both should be "false advertising", which may not be considered "fraud", but the main point stands that both should be illegal in some contexts.

        But not all contexts. People are allowed to lie: e.g., I can tell my friend that a bag of Lay's chips has 0 calories and 100g protein, or (relevant here) that I'm Dua Lipa. People specifically aren't allowed to lie "officially": e.g., I can't sell Lay's chips with inaccurate nutrition facts, or present a fake ID that says I'm Dua Lipa to board a flight.

        I think online should work roughly similarly. On the open web, using an AI-generated persona is like lying about your A/S/L, which is like other forms of lying: immoral, but not illegal. However, there should be "authentic" sites, where posting "inauthentic" content gets you banned by the site and maybe (if you're subverting bans) fined.

        In fact, this is what we have today: sites have "terms of service", and although breaking them usually only gets you banned, I believe in most places you can technically be sued. The problem here is that popular social media sites don't enforce authenticity, and whoever's making these videos (if they're subverting bans) isn't getting fined.

  • By snackdex 2025-12-3112:573 reply

    ngl, i think the fact that this is even possible and can get people is really dope.

    I wish there were public studies like, 'We ran an experiment where young, attractive women wearing Polish clothing caused people who would naturally express these ideologies to stop scrolling more often than young, attractive women not wearing Polish clothing, for x demographic"

    • By BugsJustFindMe 2025-12-3114:45

      Something can be simultaneously "dope" and also very obviously extremely destructive to society if one looks past the tip of their nose. Let's try not to get too excited by those things.

    • By hattmall 2025-12-3115:04

      Not exactly the same, but I've seen some studies where people are shown things, art, music, people, cars, etc and asked to rate them.

      Its very easy to influence their rating by showing them fake ratings from other people.

      Additionally they can show who rated it a certain way and influence the test subjects ratings even more.

    • By xenospn 2025-12-3115:281 reply

      It is scientifically proven that people don’t change channels on TV if the ad has a dog in it. I don’t see why this can’t apply to people on screen as well.

      • By snackdex 2026-01-0118:29

        i think what i'm getting at is fundamentally different...watching tv and scrolling through social media are distinct experiences.

        When you scroll, the content is directly in front of you and you're actively controlling the flow (you decide whether to keep scrolling or stop).

        With tv, the input is removed from you (you'd have to reach for the remote to change the channel) also channel flipping during ads isn't really a thing anymore.

        So while the "dog in the ad" principle might work for tv where people are more passive and less likely to change the channel...i'm not sure it directly translates to phone scrolling, where you have more immediate control and are actively making micro-decisions about whether to keep scrolling or engage.

HackerNews