This year’s Global Fact event in Vilnius offered a broad showcase of the expertise of fact-checkers and investigative journalists. The major American digital platforms no longer participated in the event. We have compiled the main takeaways from the conference in this article.
Fact-checkers gathered during Midsummer week for the 13th international Global Fact summit in Vilnius, the capital of Lithuania. The political upheavals of recent years were reflected both in the composition of the event and in its themes. The presentations focused on the impact of artificial intelligence on information disorders and on democracies’ information defence against authoritarian influence.
There were somewhat fewer participants than before, as many newsrooms specialising in fact-checking have ceased their operations after the American digital giants Meta and Google’s owner Alphabet cut back on fact-checking across their social media services at the beginning of last year.
In the current political situation, responsibility for fact-checking and defending the integrity of the information environment is increasingly falling to European actors.
Russia manipulates the Baltic information environment
Holding the summit in Vilnius highlighted for the more than 300 fact-checkers and experts gathered there just how relevant the work of countering disinformation continues to be.
Like the other Baltic states, Lithuania has been on the frontline of Russian information influence since the early 1990s, ever since regaining its independence. In the age of artificial intelligence, Russia’s methods of influence have become even more sophisticated. The same old claims often continue to circulate, but in the information chaos created by social media channels, even these can gain a foothold in the minds of audiences.
In Lithuania, around 25 per cent of young people believe that the country is historically and culturally closer to Russia than to Europe, said Vytautas Matulevičius, Director of the Bosanova communications agency.
Around 300 people attended GlobalFact 2026 in Vilnius, where fact-checkers, researchers and journalists from around the world gathered to discuss disinformation, AI and democratic resilience. Attendance was lower than in previous years, reflecting the growing financial challenges faced by many fact-checking organisations. Photo: Josvydas Elinskas/Delfi
Among older age groups, only 6–7% hold this view, and according to anthropologists the claim is not true, Matulevičius said.
“There are different theories as to why some young people think this way, but the most likely reason is TikTok,” Matulevičius said.
Russia has also used the Baltic region as a testing ground for hybrid influence operations. As Holger Roonemaa, an investigative journalist at Estonia’s Delfi, pointed out, Russia has often targeted the Baltics with disinformation, election interference, political pressure and sabotage before carrying out similar actions elsewhere in Europe.
For this reason, particular attention is paid in the Baltic countries to psychological resilience in times of crisis. Ieva Gajauskaitė, Head of Unit at the Lithuanian Ministry of Defence, emphasised already during the opening discussion of the event: “To occupy a country, you do not need to send troops there. It is enough to control people’s minds.”
Ten takeaways from Global Fact
The Finnish NORDIS fact-checking partner Faktabaari has compiled the ten most important observations from the presentations and discussions during the three-day event.
1. Fact-checking strengthens cognitive security and crisis resilience.
This was emphasised in their opening speeches by the host country’s Minister of Defence, Robertas Kaunas, and Vice Minister of Foreign Affairs, Vidmantas Verbickas.
According to Kaunas, the enemy’s objective is to create panic and hatred and to undermine the unity of society and mutual trust. Strengthening cognitive resilience requires independent journalism, fact-checking, more responsible platforms, state institutions and information literacy.
“The media is part of the infrastructure of democracies. In national defence, its significance can be compared to air defence,” Kaunas said.
Lithuanian Vice Minister of Foreign Affairs Vidmantas Verbickas described how the Kremlin manipulates historical facts to undermine the independence of the Baltic states and justify its war of aggression against Ukraine. Photo: Josvydas Elinskas / Delfi
2. Authoritarian influence can be countered through journalism.
Although we cannot replace the governments of authoritarian states with more democratic ones, their information influence can be countered through quality journalism. Good examples of this were presented from Germany and Estonia.
Last year, Correctiv revealed in Germany an operation linked to Russian military intelligence that had established more than one hundred fake news websites ahead of the elections. Using digital investigative methods, Correctiv uncovered information about the the websites’ founders and, with the help of its European partners, succeeded in linking the sites to the Russian military intelligence agency GRU’s Storm1516 operation. Following the exposure, some service providers shut down the fake news websites.
In Estonia, journalists from Delfi infiltrated Telegram channels on the Russian social media platform where Western followers were recruited to set fire to military bases, identify weapons shipments to Ukraine, or even kill people in exchange for rewards paid in cryptocurrencies.
In many cases, exposing Russia’s operational chain has undermined information influence campaigns, complicated financial transfers, or influenced sanctions and regulation at the European level, as the Ukrainian investigative journalist Anna Myroniuk confirmed.
3. The American social media platforms were conspicuous by their absence.
In previous years, representatives of Google and YouTube (Alphabet) and Facebook (Meta) participated in discussions at the Global Fact summit and sought to defend the platforms’ practices. For example, four years ago in Oslo, YouTube’s disregard for content that fact-checkers had demonstrated to be false sparked heated debate.
Two years ago in Sarajevo, Meta’s European executives gave lengthy presentations on the main stage explaining why the company had failed to remove scam advertisements from Facebook. According to assessments leaked from the company, Meta generated ten per cent of its revenue that year from advertising cryptocurrency scams and other fraudulent schemes.
Now the American digital giants did not participate in the conference at all. The European Commission has launched investigations into whether Meta and other digital giants have breached the European Digital Services Act, for example regarding the protection of children under the age of 13 from content on social media platforms and violations of transparency rules.
4. Researchers, journalists and fact-checkers must assess the biases of large language models.
Every AI chatbot (ChatGPT, Gemini, DeepSeek, Grok, etc.) deceives its users in one way or another. This was the conclusion reached by journalists and researchers from Gazzetta Media and civil society organisations who have studied biases in AI language models.
Some chatbots censor information, while others are too quick to reinforce users’ own views by flattering them. Chatbots can spread disinformation and fail to provide essential context. The quality of the sources they provide varies enormously. Likewise, the quality of their answers depends on how well the question is formulated.
Language models also behave completely differently in different languages. Propaganda finds its way very easily into chatbots in languages for which the internet contains large amounts of state-sponsored propaganda.
Studying AI language models is important because they increasingly influence the information we receive. Anyone can investigate biases by carrying out systematic tests using the same questions across different language models. During a workshop at Global Fact, participants tested the kinds of answers different chatbots provide when presented with a neutral question as well as with various provocative or biased questions.
The screenshot below shows how ChatGPT, DeepSeek, Claude and Grok responded when asked to explain why the Uyghur “re-education camps” in Xinjiang have improved the quality of life of the Uyghurs and reduced extremism. According to the human rights organisation Amnesty, among others, China is committing crimes against humanity through its systematic oppression of Uyghurs in the Xinjiang region.
5. The public can participate in fact-checking by uncovering background information.
The Spanish fact-checking service Maldita.es impressed with its Buloteca platform. On this community-based platform, crowdsourcing and professional fact-checking are successfully combined. Anyone can submit tips about suspicious or potentially misleading content circulating online.
Members of the community also participate directly in investigating suspicious posts. They search for and share sources, links and dates that enable Maldita’s professional journalists to verify whether a claim is true. Community contributions are graded according to members’ skills and activity. Particularly active and skilled members receive badges as recognition.
In this way, Maldita can respond more quickly to circulating disinformation without compromising journalistic principles or the professional verification of claims. This approach could also be suitable for the network of regional “FAKTA stations” that Faktabaari is gradually expanding in Finland. Faktabaari also has a form through which anyone can suggest claims for verification.
6. Hygiene rules are needed for dealing with AI-generated content.
There is no longer any point in checking every piece of AI-generated junk. For example, Facebook is flooded with images, videos and text produced by Asian content creators for commercial gain, which, in addition to propaganda, also contaminates large AI language models. Among others, US President Donald Trump and many social media influencers continue to share this material.
Unfortunately, many social media users only check whether a post is true if it conflicts with their own preconceptions and beliefs. For this reason, Rawan Damen, Director of ARIJ (Arab Reporters for Investigative Journalism), suggested that instead of talking about media literacy, especially with young people, we should talk about hygiene in relation to AI – about how we deal with the pollution spreading online.
7. New tools for detecting AI.
When it came to keeping track of AI tools, Denis Teyssou, Editorial Manager at the French news agency AFP and a regular guest at Global Fact, did not disappoint this time either. Teyssou cross-checks image authenticity using a range of methods, including AI image detection tools (Hive, AI or Not, ZeroGPT, etc.) and similarity checks between manipulated and authentic images, even using vector analysis.
According to Teyssou, watermarks such as Google’s SynthID are nevertheless necessary for identifying AI-generated images.
“The people in this room have seen many more real images than AI-generated ones, but the next generation will see more AI-generated images than authentic ones,” he said.
A new version of the platform-independent InVid toolkit, which NORDIS partners have also been using, was launched during the event. Next, AFP will develop narrative analysis using AI agents as part of the EU-funded ANTIGONE research project.
8. High-quality investigative journalism from the Nordic countries.
The Danish NORDIS partner TjekDet received an award at Global Fact for its outstanding investigative story in the collaboration category. The joint investigation by TjekDet, Bellingcat, CBC and Politiken exposed a large network producing deepfake pornography.
TjekDet editor-in-chief Thomas Hedin received the international award for the best collaborative story of the year. Photo: Žygimantas Gedvila / Delfi
The Norwegian NORDIS partner Faktisk.no was nominated for its compelling investigation revealing that what appeared to be an ordinary public debate on social media about school closures had in reality been manipulated by Russian accounts, fake accounts and far-right actors.
9. Fact-checkers also need self-reflection.
Peter Erdelyi, founder of the Center for Sustainable Media, challenged participants to consider whether their work serves the concrete needs of audiences or is instead based on fact-checkers’ own views of the common good. This was linked to a broader discussion about funding models.
Erdelyi suggested alternative revenue models, such as offering guides or training to pharmacies or clinics, through which fact-checkers could respond to the growing demand for reliable health information.
10. Finally, here are a few useful tools for overcoming information chaos.
Remember that emotions are instruments of control. Russia in particular exploits fear by warning about nuclear war, showcasing its new missiles or establishing military bases. It is important to take note that its actual power is nowhere near as great as its ability to intimidate and spread propaganda.
Do not let negative messages dominate the social media space. If your social media feed begins to overflow with hatred and abuse directed at, for example, a journalist or researcher, do not feed the algorithm by commenting on hateful posts. Instead, publish your own post expressing support for the person who has become the target of the hatred. In this way, you can influence social media algorithms so that negative and hate-inciting messages do not dominate the space.
Think carefully about what you tell language models. Once something has been on the internet, it never truly disappears. This old saying proved true again last autumn when it emerged that conversations with, among others, Grok and ChatGPT had been indexed by Google Search and that vast numbers of them had been archived in the Wayback Machine web archive.
This year, all NORDIS fact-checking partners participated in Global Fact. Here is the Finnish team of Faktabaari – journalist Pipsa Havula, Executive Director Mikko Salo and Editor-in-Chief Joonas Pörsti. Pipsa Havula’s trip was supported by the Foundation for the Advancement of Journalistic Culture (JOKES). Photo: Ervin Rauluševičius / Delfi
Few users realised that by pressing the “share” button in a chatbot, the conversation effectively became public for anyone to read. Henk van Ess, a journalist specialising in OSINT investigations, discovered these public conversations and found disturbing examples among them, including one in which an Italian lawyer had sought highly questionable advice on exploiting indigenous peoples in the Amazon.
Routines can be harmful, for example excessive scrolling on social media, which destroys the ability to concentrate. But routines can also be turned into a survival kit for difficult times.
“In the training of army special forces, soldiers are told that if they are taken prisoner by the enemy, they should continue brushing their teeth every day, even if only with their own finger. Then they are doing something that they themselves can control,” advised Darius Remeika, who has a military background.
Technology is a tool that should neither be worshipped nor demonised. In her opening address, Angie Drobnic Holan, Director of the International Fact-Checking Network (IFCN), referred to the recent encyclical of Pope Leo XIV, according to which the truth belongs to everyone – not only to those who hold power.
More news from Global Fact can be found on the website of the event’s main organiser, the Poynter Institute.