Impressions from the Disinfo 2026 Conference in Vilnius
This year, our Violeta Trkulja, Juliane Stiller and Joachim Griesbaum from our project team attended the Disinfo conference in Vilnius. The European conference brings together research, civil society and policymakers, and focuses on disinformation and its legal and geopolitical consequences. The full programme is available at https://disinfo2026.mitto.app/programme.
For our project DeMuKo, in which we study how people in multilingual contexts in Germany deal with information and disinformation, four aspects were particularly relevant.
AI changes manipulation on two levels
First, in one-on-one interaction: People who use chatbots or AI companions can end up in a so-called “epistemic cocoon” (Flore, 2025). The system picks up on uncertainties, asks follow-up questions and thereby reinforces them instead of putting them into context. Gradually, a closed information bubble emerges: the more users reveal, the more the responses adapt to them, and the greater the influence on their views.
Second, at scale: AI makes it easier to produce disinformation in large volumes. This shifts the balance further: creating and spreading manipulative content becomes cheaper, while verifying and debunking it remains time-consuming and costly.
For our project, this means that media literacy must also include an understanding of how AI affects the information environment as a whole, as well as how people use AI tools themselves, since these increasingly serve as a source of information. This is especially relevant when such tools are more accessible in one’s own language than, for example, German-language services.
Information deserts as a breeding ground for disinformation
This topic addressed spaces where reliable information is missing. Such voids can have two causes: either there is no information (yet) from reliable sources, or information is deliberately withheld or even suppressed by gatekeepers (including, for example, state authorities). In both cases, a gap emerges that can be filled by disinformation.
This perspective aligns with one of DeMuKo’s working assumptions: people who, for instance, have only limited access to the German-language information space in their first period after arriving in Germany often do not yet know which sources are trustworthy and have fewer ways to verify information. Such gaps can increase susceptibility to disinformation.
Diaspora communities as targets of campaigns
One aspect that came up only briefly at the conference is nonetheless of particular interest to us: one talk mentioned that disinformation campaigns also target diaspora communities, for example to maintain support for government policy in the country of origin or to influence attitudes and actions in the receiving country. Older people, for instance, may also be targeted, for example by being given access to devices that receive (banned) TV channels spreading disinformation.
When critical thinking becomes a defence
One discussion focused on the effectiveness of media literacy training and how it needs to be adapted to new challenges such as AI chatbots. Media literacy can backfire. People who strongly identify with a narrative often use the skills they have learned, such as questioning sources or looking for counter-evidence, to defend their own beliefs. Conventional training barely reaches these people.
For DeMuKo, this is a central question, as we are developing resources for learners and teachers in adult education. How media literacy can be taught without turning into a defence mechanism is a question we are taking home from Vilnius.
Key takeaways for DeMuKo:
The conference showed us that learning concepts based solely on “critical thinking” often do not achieve the desired effect. Skills associated with critical thinking are often used to defend existing beliefs. Statements that contradict one’s own views are scrutinised closely, while statements that fit one’s worldview are intuitively accepted as true. This is the well-known confirmation bias. For learning concepts, this means that concrete methods for verifying information should be taught. One promising approach is to strengthen trusted persons within language communities. They enjoy greater credibility than fact-checks alone and can therefore counter disinformation more effectively.
It is also important to analyse the so-called “sender side” as precisely and comprehensively as possible. This means mapping all channels used in each language community, from social media, IPTV, radio and messaging apps to AI chatbots. In short: rather than focusing only on the recipients of disinformation, we should also look at distribution routes and channels, paying particular attention to the needs of language communities.
References:
Flore, Massimo, Synthetic Friends: AI Companions and the Future of Disinformation (December 14, 2025). Available at http://dx.doi.org/10.2139/ssrn.5920643
Joachim Griesbaum also contributed to this post.
Comments