On January 27 and 28, the Czechs voted in the second round of the presidential elections. Petr Pavel, an independent and retired Czech army general, beat Andrej Babiš, the leader of the populist ANO and former prime minister.
The contest was marked by “the biggest disinformation campaign in Czech history”. Pavel was the target, and Babiš fanned the flames.
The Babiš campaign posted big billboards throughout the country. Their main message: Petr Pavel wants to drag the Czech Republic into war.
Citizens also received a text message saying they will receive weapons and will be mobilised, fake-signed by Pavel.
How do you fight such centralised disinformation?
In judo, to defeat an opponent, you work with their energy, not against it. In politics, this is a tactic often used by populists: they point the finger at the attacks of the centrists and make themselves the victim. Their message: “They are against me because I am for you.” But it is a tactic that can also be used by centrists.
In this spirit, Pavel’s campaign strategist Michal Repa decided to use an approach he calls “Campaign Judo”, which he developed in previous international campaigns for countering misinformation. In short, it means not to fight disinformation, but to promote it, and to make fun of it. This way, they made the attacks less relevant and could use them to positively mobilise their own voters. “Show the quantity and quality of disinfo and slander. Frame it and promote the attacks”, Repa told PartyParty.
In addition to the centralised disinformation campaign, there was also an unprecedented wave of hate and disinformation targeted at Pavel. While in the past, it was troll farms in Russia and China that did the heavy lifting in such contexts, now it was bots. How do you defeat the sheer quantity of fake content?
“Whether we like it or not: We are in the business of fighting for democracy”, Filip Strycko told PartyParty in a Zoom interview. The Slovak entrepreneur and AI expert is the co-founder of Trollwall, an AI-driven content moderation tool for Facebook and Instagram. Trollwall takes its name from the highest vertical rock face in Europe, located in Norway–a wall so steep that trolls cannot climb it.
Eastern Europe is now one of the main battlegrounds in hybrid cyberwarfare. “In hybrid warfare, scale is the key,” says Strycko, “and that scale is not going to be provided by people alone.” To change the narrative, it takes AI to fight AI.
The Pavel campaign decided to use Trollwall to hide hate and disinformation on their social media channels. The tool hid tens of thousands (!) of slanderous and hateful comments each month, a whopping 25% of the total.
For the tool to be effective, the specific language is crucial, says Strycko: “The model has to be trained to master specific national phrases.” After all, a slander in Czech might not make sense in English, and vice versa.
The Pavel campaign demonstrated how disinformation can be successfully countered on a large scale. Their win serves as an important lesson for those who value democracy. If they can do it, others can do it, too.