TRUST-AI 2025
The European Workshop on Trustworthy AI 2025

Table of Contents

• Preface
  Summary: There were 60 papers submitted for peer-review to this workshop. Out of these, 37 papers were accepted for this volume, all short papers.

Fairness, Diversity, and Bias-mitigation (Plenary Session 1)

• Diverse and Private Synthetic Datasets Generation for RAG evaluation: A Multi-agent Framework 1-9
  Ilias Driouich, Hongliu Cao, Eoin Thomas
• Automated and Augmented Evaluation of Bias in LLMs for High- and Low-Resource Languages 10-18
  Alessio Buscemi, Cédric Lothritz, Sergio Morales García, Marcos Gomez-Vazquez, Robert Clarisó Viladrosa, Jordi Cabot, German Castignani
• Learning Fairer Representations with FairVIC 19-29
  Charmaine Barker, Daniel Bethell, Dimitar Kazakov
• Assessing the Fairness of AI Systems for Education 30-38
  Velislava Hillman, Katarzyna Barud, Ibrahim Sabra, Clara Saillant, Syed Zulkifil Haider Shah, Lukas Faymann, Edoardo Pareti, Leo Bianchi, Manuele Barbieri

Enhancing AI Trustworthiness (Plenary Session 2)

• Pollution with Purpose: The Role of Data Quality in Trustworthy AI 39-49
  Leonie Louisa Etzold, Tim Robin Kosack, Oscar Hernán Ramírez-Agudelo, Clemens Danda, Michael Karl
• Cross-Layer Attention Probing for Fine-Grained Hallucination Detection 50-60
  Malavika Suresh, Rahaf Aljundi, Ikechukwu Nkisi-Orji, Nirmalie Wiratunga
• Explaining Concept Drift via Neuro-Symbolic Rules 61-72
  Pietro Basci, Salvatore Greco, Francesco Manigrasso, Tania Cerquitelli, Lia Morra
• Trustworthiness-as-Reward: Improving LLM Performance on Text Classification through Reinforcement Learning 73-80
  Yiqing Zhao, Xiaohui Shen, Lanfeng Pan
• Intersectional Fairness in Healthcare AI: A Pipeline-Wide Evaluation of Multi-Stage Mitigation Strategies 81-90
  Shane Kennedy, Michael Farayola, Daniel Kelly, Irina Tal, Takfarinas Saber, Regina Connolly, Malika Bendechache

Trust in Socio-technical AI Systems (Plenary Session 3)

• Trust in Vision-Language Models: Insights from a Participatory User Workshop 91-102
  Agnese Chiatti, Lara Piccolo, Sara Bernardini, Matteo Matteucci, Viola Schiaffonati
• Towards Trustworthy AI in STEM Education: Challenges and Strategies from the Trust-AI Platform 103-111
  Nikolaos Antonios Grammatikos, Evangelia Anagnostopoulou, Dimitris Apostolou, Gregoris Mentzas
• Rethinking Trust in Responsible AI 112-119
  Marina Tropmann-Frick, Michael Gille, Susanne Draheim, Philine Pommerencke, Maximilian Kiener, Jonas Bozenhard
• Can I Trust My Trajectory Prediction Model? 120-128
  Franz Motzkus, Christian Schlauch, Sebastian Bernhard, Ute Schmid

Measurement and Assessment of Trustworthy AI (Plenary Session 4)

• Coverage of LLM Trustworthiness Metrics in the Current Tool Landscape 129-141
  Lennard Helmer, Benny Stein, Tim Ufer, Elanton Fernandes, Hammam Abdelwahab, Abhinav Pareek, Joshua Woll
• TAI Scan Tool: A RAG-Based Tool with Minimalistic Input for Trustworthy AI Self-Assessment 142-150
  Athanasios Davvetas, Xenia Ziouvelou, Ypatia Dami, Alexios Kaponis, Konstantina Giouvanopoulou, Michael Papademas
• Towards the Assessment of Trustworthy AI: A Catalog-Based Approach 151-159
  Marco Anisetti, Claudio Agostino Ardagna, Nicola Bena, Aneela Nasim
• Patterns for Semi-Automated Trustworthiness Risk Assessment of AI Systems in Cyber-Physical Environments 160-168
  Samuel Senior, Steve Taylor
• An Entropic Metric for Measuring Calibration of Machine Learning Models 169-179
  Daniel James Sumler, Lee Devlin, Simon Maskell, Richard Lane

AI Compliance and Explainability (Plenary Session 5)

• XAI Desiderata for Trustworthy AI: Insights from the AI Act 180-187
  Martin Krutský, Jiří Němeček, Jakub Peleška, Paula Gürtler, Gustav Šír
• Towards Abductive Latent Explanations 188-199
  Jules Soria, Zakaria Chihani, Julien Girard-Satabin, Alban Grastien, Romain Xu-Darme, Daniela Cancila
• Topic Modelling on the European AI Act 200-208
  Ettore Carbone, Alex Giulio Berton, Purbasha Chowdhury, Teresa Scantamburlo, Paolo Falcarin
• Lightweight AI Governance (LAIG) Framework for SMEs 209-215
  Aleksander Młodawski, Aleksandra Wolniak

Enhancing Performance and Integrity (Poster Sessions)

• Discriminator-Guided Unlearning: A Framework for Selective Forgetting in Conditional GANs 216-224
  Byeongcheon Lee, Sangmin Kim, Sungwoo Park, Seungmin Rho, Mi Young Lee
• A Comparison of Human and Machine Learning Errors in Face Recognition 225-236
  Marina Estévez-Almenzar, Ricardo Baeza-Yates, Carlos Castillo
• Artificial Conversations, Real Results: Fostering Language Detection with Synthetic Data 237-246
  Fatemeh Mohammadi, Tommaso Romano, Samira Maghool, Paolo Ceravolo
• SP-Guard: Selective Prompt-adaptive Guidance for Safe Text-to-Image Generation 247-255
  Sumin Yu, Taesup Moon
• Transductive Model Selection under Prior Probability Shift 256-265
  Lorenzo Volpi, Alejandro Moreo, Fabrizio Sebastiani

Reliability and Interpretability (Poster Sessions)

• Balancing Accuracy and Interpretability in Multi-Sensor Fusion through Dynamic Bayesian Networks 266-275
  Franca Corradini, Carlo Grigioni, Alessandro Antonucci, Jerome Guzzi, Francesco Flammini
• Beyond Single-model XAI: Aggregating Multi-model Explanations for Enhanced Trustworthiness 276-286
  Ilaria Vascotto, Alex Rodriguez, Alessandro Bonaita, Luca Bortolussi
• How to Build Trust in AI Systems with Misclassification Detectors and Local Misclassification Explorations 287-297
  Pål Vegard Bun Johnsen, Milan De Cauwer, Joel Bjervig, Brian Elvesæter
• Multi‑Domain Calibration Framework for SAR‑XAI: A Systematic Approach to Trustworthy Explainable AI with Transparency Enhancements 298-306
  Diego Argüello Ron, Christyan Cruz Ulloa, Orfeas Menis Mastromichalakis, Kristina Livitckaia, Jaime Del Cerro, Oscar Garcia Perales, Pawel Andrzej Herman
• Quantifying Dataset Trustworthiness from Labeling Bias Using Subjective Logic 307-314
  Koffi Ismael Ouattara, Ioannis Krontiris, Theo Dimitrakos, Frank Kargl

Philosophical and Societal Considerations (Poster Sessions)

• Do Foundation Models Learn Fair Representations? A Critical Evaluation of TabPFN on Algorithmic Fairness Benchmarks 315-323
  Sam Schiffman
• Fair Enough? A Map of the Current Limitations of the Requirements to Have Fair Algorithms 324-336
  Daniele Regoli, Alessandro Castelnovo, Nicole Inverardi, Gabriele Nanino, Ilaria Penco
• Monitoring Historical Cultural Hacking in Large Language Models 337-345
  Fabio Celli, Astik Samal
• EAIIM - Ethical AI Impact Matrix for High-Stakes Decision Systems Framework 346-357
  Pedro Oliveira, Tomás Francisco, Pedro Oliveira, Manuel Rodrigues
• Fictionalism about Agentic AI 358-366
  Anthony R.J. Fisher