Planned Program – September 22, 2026

The workshop is part of the Informatikfest of the German Informatics Society (GI) in Dresden.

📍 Dresden, Germany, Room BAR/I88/U
📅 Tuesday, September 22, 2026
🌍 Workshop language: English
💻 Hybrid participation planned

Here you find the Program at a glance

Disinformation Space

From Prompt to Propaganda: Generative AI and the New Disinformation Lifecycle

What to Expect

Generative AI is fundamentally changing how information is created, manipulated, and propagated. The barrier to producing convincing misinformation and disinformation has never been lower, while the scale and speed of influence operations continue to grow.

Understanding these developments requires much more than isolated technical solutions. It demands collaboration across computer science, cybersecurity, media studies, linguistics, psychology, political science, law, industry, government, journalism, and civil society.

Disinformation Space is therefore not designed as a conventional sequence of research talks.

We want to bring these communities together for a full day of research, operational experience, experimentation, debate, and collaboration.

The day has three overarching goals:

🗺️ Map the ecosystem. Identify the actors, initiatives, projects, and competencies currently shaping research and practice in Germany, France, Ukraine, and Europe.

🧭 Develop a shared roadmap. Discuss research priorities, capability gaps, collaboration opportunities, and what Europe needs to address next.

🤝 Shape a common understanding. Work towards an operational understanding of disinformation and misinformation that can connect research, industry, policy, media, and civil society.

And we do not want participants simply to listen.

During the afternoon, you will first test whether you can recognize manipulated information, then create it yourself, and finally see whether you can fool the room.

All scientific contributions will additionally be presented as posters during the coffee breaks, giving participants and authors considerably more time for discussion than the short presentation slots alone allow.

🌍 Morning – Learning from National Ecosystems

How do different countries prepare for and respond to information manipulation?

The morning brings together three deliberately different perspectives: the German research ecosystem, the French operational ecosystem, and lessons from Ukraine under conditions of ongoing information warfare.

We then bring these perspectives together to ask a larger question:

What should a European disinformation ecosystem look like?

09:00 – 09:10 | 👋 Welcome and Opening

From Prompt to Propaganda: Generative AI and the New Disinformation Lifecycle

We open the day by introducing the lifecycle perspective behind the workshop and the questions we want to answer together.

Where does disinformation begin? How is generative AI changing its creation and propagation? Where can we intervene? And how can fragmented research and operational communities work together?

09:10 – 09:45 | 📰 Opening Keynote – The Media Perspective

Caroline Lindekamp, CORRECTIV – to be confirmed

We begin outside the traditional technical perspective.

How is the information ecosystem changing from the perspective of journalism and fact-checking? What happens when social media, traditional media, platform dynamics, and generative AI interact? And what does this mean for those trying to establish what is true?

The keynote sets the stage for a day that deliberately connects technology with the societal environments in which information actually circulates.

🇩🇪 Germany – Building a National Research Ecosystem

09:45 – 10:30

From Individual Projects to a Connected Research Landscape

Maja Denisova and representatives of German projects and initiatives

Germany has numerous research activities addressing different dimensions of misinformation and disinformation.

But who is doing what? Where are the competencies? Where do projects overlap? And where are important gaps?

Representatives of German initiatives will provide concise insights into their activities, approaches, and results.

Rather than presenting projects in isolation, the objective is to start assembling a picture of the German disinformation research ecosystem and its potential connections to European partners.

The conversation continues with the project and scientific posters during the coffee break.

10:30 – 11:00 | ☕ Coffee Break and Poster Session

Meet the People Behind the Research

This is deliberately more than a coffee break.

Scientific contributions and ecosystem initiatives will be available as posters, giving participants the opportunity to go deeper into methods, results, projects, and potential collaborations.

Use the time to find people working on your problem, challenge the authors, compare approaches, and discover potential partners.

🇫🇷 France – From Research to Operational Capabilities

11:00 – 11:20

Connecting Cybersecurity, Influence Operations, Industry, and Public Institutions

Jean-Luc Gibernon, Pole d’Excellence Cyber / Sopra Steria, and further contributors

The French perspective adds a stronger operational dimension.

How can research, cybersecurity expertise, industry, public institutions, and operational communities work together against information manipulation?

The session will particularly explore activities around the Pole d’Excellence Cyber and the emerging opportunities for stronger Franco-German and European collaboration in the domain of information influence.

11:20 – 11:35 | 🏙️ When Disinformation Reaches the Real World

AI-Generated Disinformation in Times of Urban Crisis: Lessons from the Rival Youth Groups Case in North Holland

Sandrine Turgis and Leandre Lebon

What happens when manipulated information leaves the online environment and starts affecting schools, municipalities, public anxiety, and crisis management?

This contribution examines a concrete urban crisis involving rival youth groups and shows how AI-generated and manipulated information can create consequences far beyond social media.

The case brings law, political science, public administration, and AI-enabled disinformation into the same discussion.

10-minute presentation + 5-minute discussion

🇺🇦 Ukraine – Lessons from the Frontline

11:35 – 12:00

What Changes When Information Warfare Is Not a Scenario?

Anatolii Marushchak – to be confirmed

The morning progresses from research and operational preparedness towards experience under actual conflict conditions.

Ukraine has been continuously exposed to sophisticated information operations as part of a wider hybrid conflict.

What has worked? What has failed? Which assumptions made in research survive contact with operational reality? And what should the rest of Europe learn from Ukraine’s experience?

This session brings a frontline perspective to the questions raised throughout the morning.

🇪🇺 From National Ecosystems to European Action

12:00 – 12:10 | ⚖️ Opening Position – Is Technology Enough?

More Regulation, Please: Technical Privacy Protection Alone Can Only Partially Limit Democratic Harm from Microtargeting – A Position Statement

Peter Schoo

After discussing research capabilities, operational approaches, and frontline experience, we deliberately broaden the question:

Can technology alone solve the problem?

Peter Schoo argues that technical privacy protection can address only part of the democratic harm created by microtargeting and opens the discussion towards governance and regulation.

His position statement serves as the provocation for the following European panel.

10-minute presentation followed by discussion within the panel

12:10 – 12:30 | 💬 Panel Discussion

Building the European Disinformation Ecosystem

Panelists

Maja Denisova – German research ecosystem
Jean-Luc Gibernon – French operational and cybersecurity ecosystem
Anatolii Marushchak – Ukrainian frontline perspective
Peter Schoo – Governance and regulation

Research. Operations. Frontline experience. Regulation.

The four perspectives come together to ask not simply what individual countries are doing, but what Europe should do next.

Among the questions we want to discuss:

Where are the strengths of the different ecosystems?
What can they learn from one another?
What capabilities and actors are still missing?
Where can technology help, and where are technical solutions insufficient?
What can research and policy learn from operational and frontline experience?
How can European collaboration be strengthened?
What should the common research and innovation agenda be?
How do we move from individual projects towards an integrated European ecosystem?

The discussion feeds directly into the workshop’s ecosystem mapping, roadmap, and common understanding of disinformation.

12:30 – 14:00 | 🍽️ Lunch Break

🧪 Afternoon – Experiencing the Disinformation Lifecycle

The afternoon changes perspective.

We stop merely talking about disinformation and start experimenting with it.

First, you will try to recognize manipulation.

Then you will become the creator.

Next, researchers will show how AI can turn individual manipulation into scalable infrastructure and propagation.

Finally, we will put the content created by participants to the test in front of the entire room.

Part I – 🎭 Creating Influence

14:00 – 15:00 | Hands-on Challenge

Can You Spot It? Can You Create It?

How good are you at recognizing manipulated information?

And once you understand the challenge, can you create something convincing yourself?

🔎 Step 1 – Can You Tell What Is Real?

Participants begin with ten rounds of JudgeGPT.

Real or manipulated?

Make your decision, test your intuition, and discover which signals you rely on when judging information.

How confident are you? Which cues make something look trustworthy? And are those cues actually reliable?

🛠️ Step 2 – Now Switch Sides

Participants form small teams and select a current real news story from a credible news website.

The challenge:

Turn it into convincing misinformation or disinformation.

But simply inventing something false is not enough.

Each team first has to develop an influence strategy:

🎯 Goal – What should your audience believe, feel, or do?

👥 Target audience – Who exactly are you trying to influence?

🧠 Technique – What will you omit, exaggerate, reframe, decontextualize, or invent?

📖 Narrative – Why should your version appear plausible?

🤖 Execution – How can generative AI help you create or refine it?

Teams retain both the original news story and their manipulated version.

Three teams will later put their creations to the ultimate test:

Can you fool the workshop audience?

Part II – ⚙️ From Infrastructure to Propagation

15:00 – 15:30

You have just created one piece of manipulated information.

But what happens when AI makes this process automated, multilingual, coordinated, and scalable?

Two complementary research contributions follow the transition from creation to systemic propagation.

15:00 – 15:10 | 🤖 Building Influence

[GenTool] as Agentic Infrastructure: Seeding Cross-Lingual Misinformation Simulations and Red-Teaming Social Media Agents

Alexander Loth, Martin Kappes, Marc-Oliver Pahl

Generative AI is evolving from a tool that produces individual pieces of content into an infrastructure capable of supporting increasingly autonomous influence operations.

What happens when agents can generate, adapt, translate, and seed content across different environments?

10-minute presentation

15:10 – 15:20 | 🌐 Spreading Influence

From Peripheral Seeding to Systemic Contamination: Multi-Vector Monitoring of Information Propagation

Jordan Ricker, Clement Benesse, Darius Beucler

Creating a narrative does not make it influential.

This contribution examines how narratives move from peripheral actors into wider information ecosystems through coordination, amplification, and cross-platform propagation.

10-minute presentation

15:20 – 15:30 | 💬 Joint Discussion

From Generation to Systemic Propagation

What changes when influence becomes scalable?

The authors and audience connect the two perspectives and discuss the transition from AI-enabled generation to coordinated propagation across platforms and communities.

15:30 – 16:30 | ☕ Scientific Poster Session and Networking

One Hour to Go Deeper

All scientific contributions will be available as posters.

The extended break provides time for the discussions that cannot fit into short presentation slots:

challenge methodologies,
explore technical details,
discuss applications,
meet potential collaborators,
connect scientific and operational perspectives,
and continue building the ecosystem map started during the morning.

Part III – 🗳️ Can We Still Tell?

16:30 – 17:15 | Live Challenge, Reveal, Discussion, and Expert Survey

The experiment started at 14:00 now comes full circle.

You have tried to recognize misinformation.

You have created it.

You have heard how it can be industrialized and propagated.

Now we test the room.

16:30 – 16:55 | 🎭 The Live Challenge – Real or Fake?

Three teams take the stage.

Each presents:

one real news story + one manipulated version.

But they do not tell you which is which.

The audience votes live:

Which one is real?

After each vote comes the reveal.

The team then takes us behind the scenes:

What was their objective?
Who was their target audience?
Which manipulation techniques did they choose?
What did they omit, exaggerate, invent, or reframe?
How did they use generative AI?
Why did they expect the manipulation to work?

And, most importantly:

Did it work?

Together we examine what fooled the audience, what triggered suspicion, and which signals people relied upon.

The goal is not to produce the most spectacular fake.

It is to understand how little may need to change before truthful information becomes persuasive manipulation.

16:55 – 17:15 | 🧭 From Experience to Evidence – Expert Survey

Now we turn the experience of the day into input for the future.

Participants contribute their expertise through a structured survey addressing questions such as:

What should we actually consider disinformation?
Where are the most important gaps in current research?
Which defensive capabilities are missing?
Which stages of the disinformation lifecycle require the most urgent attention?
Which technical, societal, educational, legal, and regulatory responses appear most promising?
Which actors and competencies already exist?
Where should European collaboration focus next?

Your answers contribute directly to the three workshop objectives:

🗺️ Mapping the ecosystem

🧭 Developing a shared roadmap

🤝 Shaping a common understanding of disinformation

Part IV – 🧠 Understanding Human Vulnerability

17:15 – 17:30

Interrupting the Chain: Human Perception of AI-Generated Disinformation Through a Kill Chain Lens

Alexander Loth, Martin Kappes, Marc-Oliver Pahl

You have just experienced how difficult it can be to distinguish real from manipulated information.

Now we put that experience into scientific perspective.

How do humans perceive AI-generated disinformation? Where are we vulnerable along the disinformation kill chain? And at which points can interventions effectively interrupt it?

10-minute presentation + 5-minute discussion

Part V – 🔬 Understanding and Detecting Manipulation

17:30 – 18:00

The final scientific session asks two complementary questions:

What makes language manipulative?

And:

Can humans and AI reliably recognize manipulation?

17:30 – 17:40 | 🔎 Detecting Influence

Not All Errors are Equal: Comparing Humans and LLMs on Detecting Argumentative Flaws in Misinformation

Maged Mortaga, Marvin Sieger, Moritz Goeke, Andre Calero Valdez

Do humans and large language models recognize the same weaknesses in misinformation?

This contribution compares their ability to identify argumentative flaws and asks what their differences can teach us about future human-AI approaches to misinformation detection.

10-minute presentation

17:40 – 17:50 | 🗣️ Language as a Weapon

Scalar Evaluative Expressions and Their Manipulative Use in LLM-Generated Text

Maurice Langner

Manipulation does not always require fabricated facts.

Sometimes seemingly subtle linguistic choices are enough to change how information is perceived.

This contribution examines evaluative expressions and their role in the persuasive and manipulative potential of LLM-generated communication.

10-minute presentation

17:50 – 18:00 | 💬 Joint Discussion and Closing

What Makes Manipulation Work – and What Do We Do Next?

The final discussion connects the scientific contributions back to everything participants have experienced throughout the day.

What made the manipulated stories persuasive?

What could humans detect?

What could AI detect?

Where should we intervene?

And what have we learned about the broader ecosystem required to address the problem?

We close by returning to the three questions that guided the workshop:

🗺️ Who is already working on the problem, and what is missing?

🧭 Where should research, policy, and collaboration go next?

🤝 Can we develop a common understanding of disinformation that allows different communities to work together?

The workshop ends at 18:00, but we hope the discussion and collaborations will continue well beyond Dresden.