For the third year running, ECAT is hosting a workshop bringing together researchers to discuss their latest insights into the functioning and impact of some of the most widespread technologies in Europe.
Engaging with independent researchers and understanding different perspectives on DSA-relevant research debates is central to ECAT’s mission, which includes providing technical and scientific insights for the enforcement of the Digital Services Act (DSA). The ECAT Research Workshop offers a whole day of focused discussions, bringing scientific insights directly into the orbit of policymakers.
This year, the event is spotlighting the emerging impact of generative AI on how online platforms are designed and experienced. Experts will focus on how the application of this rapidly evolving technology may introduce or exacerbate risks, and explore whether it offers new opportunities for risk mitigation.
The event will offer a chance for researchers to come together to learn from each other, exchange ideas and consider future collaborations. We will hear from experts through keynotes, panel discussions and poster sessions displaying the work of researchers whose proposals are accepted through our Call for Contributions.
Think this sounds interesting and relevant to your work? You can join the waiting list below to express your interest in joining. If your spot is confirmed, we will get in touch with you to provide the full details, including exact location
Please note that the event is primarily targeted at researchers, and attendance is subject to approval. For attendees who are not joining in a speaking capacity, we are unfortunately unable to reimburse travel costs.
As the event is currently at full capacity, we are operating a waiting list. However, as we may receive cancellations, please do still sign up if you are interested in joining.
Please note that the event agenda is subject to changes.
- Wednesday 25 November 2026, 09:00 - 17:30 (CET)
Programme
- 25 Nov 2026, 09:00 - 09:10 (CET)Welcome
Francesca Campolongo, Director of Digital Transformation, AI and Data, Joint Research Centre, European Commission
- 09:10 - 09:20 (CET)Opening address
- 09:20 - 09:30 (CET)ECAT introduction
Yves Punie, Acting Head of Unit, European Centre for Algorithmic Transparency, Joint Research Centre
- 09:30 - 10:30 (CET)How is generative AI reshaping social media platforms? From social interactions to synthetic, personalised and hybrid environments
Social media platforms are integrating generative AI across a growing range of functions, from content creation and editing tools to conversational agents, recommendations, advertising and moderation support. These features often appear fragmented and experimental, yet even in this relatively early phase, they may already be changing what users see, how they participate, and what “social media” is for.
The session will examine how genAI is altering social media so far and its possible impact on the DSA compliance and enforcement. The panel will discuss which changes are most visible and consequential, how they may alter the risk landscape, and how researchers and regulators can better understand and prepare for future developments.
Moreover, they will look at which features of social media may remain distinctly “social”, if users increasingly rely on genAI for interaction and content.
Panellists:
- Luca Maria Aiello, Professor, ITU Copenhagen
- Natalia Stanusch, Researcher, AI Forensics and PhD candidate, University of Amsterdam
- Stratis Tsirtis, Postdoctoral Researcher, Hasso Plattner Institute
- 10:30 - 11:00 (CET)Poster 'lightning round' presentations
During this session, the following successful submissions to the workshop’s Call for Contributions will be presented briefly:
- "If I were buying for your requirements, I'd choose the Sony WF-C710N": An audit of AI-generated commercial advice, presented by Lucas G. Uberti-Bona Marin
- Legal and Technical Shortcomings of DSA Transparency for Cyber Violence Against Women: A Consent-Based Notice Schema, presented by Raquel De Haro i Perez
- Recommending Music in the Age of Generative AI, presented by Lorenzo Porcaro
- Retouched advertisement and the rise of AI in marketing: A multi-stakeholder perspective, presented by Clara Julia Reich
- The Generative Turn in Social Media: Rethinking Personalization, Trust, and Memory over Time, presented by Alexandra Weilenmann/Alan Said
- When the machine says “Yes”: a Scoping Review mapping AI Sycophancy in LLMs, presented by Alice Barsanti
Event participants will have the opportunity to explore the insights further in the following dedicated poster sessions.
- 11:00 - 11:30 (CET)Poster session (coffee break)
- 11:30 - 12:45 (CET)The impact of genAI on cognition, emotions and how to protect vulnerable audiences
The diffusion of genAI-powered chatbots across digital services has rapidly increased their user base and reshaped the nature of many of our interactions. In this panel, we take a closer look at how generative AI may be changing human behaviour, and perhaps even our way of thinking, feeling and paying attention.
In particular, panellists will discuss how the ready availability of generative AI can impact users who are cognitively, emotionally or socially vulnerable, such as children.
Protecting such groups is an important objective of the DSA and wider regulatory obligations, and the conversation will focus on which AI features may reduce or increase risks of user manipulation and adverse mental health consequences.
Looking ahead, speakers will assess what actions can be taken by developers, regulators and users/guardians to ensure generative AI enhances rather than diminishes the human experience.
Panellists:
- Femke Konings, Postdoctoral Researcher, KU Leuven
- Carles Soriano-Mas, Associate professor, University of Barcelona
- Jessica Szcuka, Research Group Leader, University of Duisburg-Essen
- René van Bavel, Senior Researcher, Joint Research Centre, European Commission
- 12:45 - 14:00 (CET)Lunch
- 14:00 - 14:30 (CET)Searching for information in the genAI era
Commercial LLMs are evolving from text generators into complex systems that combine generation with web search, external data and other tools. At the same time, conversational systems are becoming an important interface for accessing online information. This raises questions about how they shape information exposure, especially relevant with the recent designation of ChatGPT as a very large online search engine under the DSA.
In this talk, the speaker will present outcomes from a systematic comparison of LLM-based web search with traditional search engines across 24 sociopolitical topics. The audit reveals substantial differences in source diversity, the distribution of attention across sources and the ideological orientation and factuality of the sources surfaced.
The longer-term interaction between users and AI-based recommendation systems will also be addressed. The talk will examine how potential feedback loops can alter exposure, consumption diversity and collective behaviour over time.This research raises broader questions about how emerging technologies may reshape online information ecosystems and why their effects deserve further research and monitoring, particularly in relation to the systemic risks identified by the DSA.
Speaker:
- Marco Minici, Researcher, ICAR-CNR
- 14:30 - 15:15 (CET)Poster 'lightning round' presentations
During this session, the following successful submissions to the workshop’s Call for Contributions will be presented briefly:
- A Code-Level Audit of X’s Recommendation Pipeline and Its Implications for Systemic Risk Auditing, presented by Luka Bekavac
- AI-Generated Political Content and Platform Enforcement: Evidence from Hungary's 2026 Parliamentary Elections, presented by Daniela Alvarado Rincòn
- Auditing Exposure to Harmful Content on TikTok using Multimodal Language Models: A Cross-National, Age-Stratified Study, presented by Francesco Pierri
- Blind to Harm: Risks and possibilities of AI-based content moderation on pornographic VLOPs, presented by Alessia Tranchese
- Capturing Risk Pathways in Generative AI: Lessons from Recommender Systems and Children, presented by Robin Ungruh
- Gender-Based Cyberviolence and Platform Governance: Empirical Insights from the EU Digital Services Act, presented by Adriana Mutu
- Practices Before Principles: Regulating Systemic Risks from Generative AI under the DSA, presented by John Albert
- The Road to the Ballot: a citizen‑centred framework for generative AI systemic risks to elections under DSA article 34, presented by Konrad Kiljan
- Using Child Anti-Personas to Understand the Implications of Systemic Risks, presented by Diletta Micol Tobia
- Where AI Security Meets Platform Risk: Mapping Generative-AI Threat Taxonomies onto DSA Article 34 Systemic Risk Assessment, presented by Corina Pascu
Event participants will have the opportunity to explore the insights further in the following dedicated poster session.
- 15:15 - 15:45 (CET)Poster session (coffee break)
- 15:45 - 16:15 (CET)Evaluating GPAI models for harmful manipulation
As GPAI models becomes integrated into platforms and search engines, how can we assess risks of harmful manipulation? This presentation introduces new evaluations for harmful manipulation and applies it to different models, examining sycophancy, persuasion, influence and deception, alongside custom evaluations in mental health and political contexts.
Speaker:
- Carlos Mougan, AI Office, DG CNECT, European Commission
- 16:15 - 17:15 (CET)genAI and the fight for information integrity
When generative AI became widely available, one of the first major concerns was its application in production and mass distribution of mis- or disinformation. As the quality of AI-generated material has only continued to increase, spotting what is real and what is fake has become harder, especially to the untrained eye.
Enhanced opportunities for such deception have consequences for civic discourse and potentially for electoral integrity, both highlighted as systemic risks in the DSA and an important focus of its compliance and enforcement. Through this discussion, we will learn more about the concrete impact of genAI on the information integrity environment to date. The panel will address questions such as the extent to which the quality of genAI material has changed the levels of persuasiveness of mis/disinformation, if the so-called ‘liar’s dividend’ has increased, and how AI content is being deployed by political actors for propaganda purposes.
Lastly, experts will address what platforms can do to address the risks of amplifying material that does not show what it claims to show.
Panellists:
- Michal Klincewicz, Assistant Professor, Tilburg University
- Carlos Hernández-Echevarría, Assistant Director, Maldita
- 17:15 - 17:30 (CET)Closing address
Speakers
Practical information
- When
- Wednesday 25 November 2026, 09:00 - 17:30 (CET)
- Languages
- English