Welcome to FEAR 2026

The Third International Workshop on Formal Ethical Agents and Robots (FEAR) will be held on 3 November 2026 with co-located tutorial scheduled for 2 November 2026 both at the Kilburn Building, the University of Manchester, UK.

Overview

Recent advances in artificial intelligence have led to a range of concerns about the ethical impact of the technology. This includes concerns about the day-to-day behaviour of robotic systems that will interact with humans in workplaces, homes and hospitals. One of the themes of these concerns is the need for such systems to take ethics into account when reasoning. This has generated new interest in how we can specify, implement and validate ethical reasoning. The aim of this workshop would be to look at formal approaches to these questions. Topics of interest include but are not limited to:

Schedule and Location

This year we will hold a free one-day workshop with informal proceedings, consisting of short talks and opportunities for discussion, and a free one-day tutorial that engages conversation in the state-of-the-art machine ethics fields. The workshop and tutorial will be held at Atlas room, the first floor of the Kilburn Building, the University of Manchester on the Oxford Road, M13 9PL. The best way to enter and leave the building is through the entrance on the North side of the building. See the map of the first floor below.

FEAR will be a hybrid meeting, supporting both virtual and in-person attendance. We strongly encourage in-person attendance. Registration is free, but required via this link.

Tutorial Schedule 2nd of November 2026
11:00Welcome
11:15Ethics of Human and AI - Malcolm Brown, University of Bath, United Kingdom
12:15Lunch
13:00Ethics Under Uncertainty - Marcus Pivato, Université Paris 1 Panthéon-Sorbonne
14:00The SLEEC Framework for Normative Requirements Engineering - Radu Calinescu, University of York, United Kingdom
15:00Coffee break
15:30Policy, Law, Norms and Artificial Intelligence - Emma Carmel, University of Bath, United Kingdom
16:30Closing
Workshop Schedule 3rd of November 2026
10:00Welcome
10:15Keynote Talk: Resilient Normative Alignment for Autonomous Agents - Professor Radu Calinescu, University of York, United Kingdom
11:15Interpretable Moral Decision-Making under Epistemic Uncertainty - Jessica Ciupa, Vaishak Belle, Antonia Sewell
Practical Reasons for Artificial Agents: From Reason-Constrained to Reason-Responsive (PO)MDPs - Federico Faroldi
Metanormative Theory for RL-Based Moral Agents - Aleks Knoks, Marija Slavkovik
Modeling Moral Evaluation Systems in Normative Moral Philosophy for Machine Ethics - Guillaume Gervois, Gauvain Bourgne, Marie-Jeanne Lesot
11:40Lunch + Poster Session
13:00Keynote Talk: Developing and Applying Explainable AI Using Computational Models of Argument - Professor Katie Atkinson, University of Liverpool, United Kingdom
14:00Bringing ALOn to Life - Bijan Parsia, Uli Sattler
14:15Runtime Monitoring and Mitigation for Algorithmic Bias - Karen Onwuazo, Raynaldio Limarga, Louise A. Dennis
Normative Quiescence after Agent Cancellation - Avinash Goutham Aluguvelly
Who Sees to It That AI Complies? Towards a Choice-Based STIT-Deontic Semantics for Accountability under the EU AI Act - Andrea Vestrucci
Institutionalised Memory for Theory of Mind in LLM Agents: A Deontic Framework for Belief Layering - Ambrose Brown, Jessica Ciupa, Vaishak Belle
14:40Coffee break + Poster Session
15:10Community Discussion
16:00Closing

Keynote Speaker

Resilient Normative Alignment for Autonomous Agents
Professor Radu Calinescu
University of York, United Kingdom
Abstract

Starting from a recently introduced formal framework for operationalising social, legal, ethical, empathetic and cultural (SLEEC) norms for autonomous agents, this keynote asks how we can ensure that normative compliance is maintained when the agents face real-world uncertainty. Two approaches to addressing different aspects of this challenge will be presented. The former approach, SLEEC-ADAPT, uses dynamic workflow reconfiguration to adapt the behaviour of an agent at runtime as it encounters users with different SLEEC requirements. The latter, SLEEC-FIASCO, is a design-time methodology that uses a combination of formal analysis and generative AI to systematically determine SLEEC-relevant disruptions and associated mitigations for an autonomous agent under development. Together, they provide an initial foundation for achieving resilient normative alignment in real-world deployments.

Radu Calinescu
Radu Calinescu
Biography

Radu Calinescu is Professor of Computer Science at the University of York, UK, where he leads a research team developing theory and tools for the modelling, verification, synthesis, and assurance of self-adaptive, autonomous, and AI systems. A central focus of his recent work is the operationalisation of normative requirements for autonomous agents. He has published over 200 research papers in the areas of software engineering, formal methods, and probabilistic modelling. He has served as a Program Co-Chair for international conferences including ICECCS, SEFM, SEAMS, ICSA, and ICPE, and is an Associate Editor for ACM Computing Surveys and ACM Transactions on Autonomous and Adaptive Systems.

Developing and Applying Explainable AI Using Computational Models of Argument
Professor Katie Atkinson
University of Liverpool, United Kingdom
Abstract

With AI-based technology now becoming firmly deployed in a range of domain applications, questions around scrutability, explainability and trustworthiness of AI tools used in high-stakes domains are rightly receiving more attention. In this talk, I will demonstrate how techniques from the field of computational models of argument are being used to develop decision support tools that are able to capture domain reasoning that has strong fidelity to how humans reason in these domains, particularly where there is conflicting information that needs to be considered and resolved. I will provide an insight into some specific techniques developed within the field of computational models of argument and their application in two very different domains: law and chemistry. I will also outline how these symbolic AI methods are being used together with data-driven AI techniques to produce hybrid approaches that draw upon the benefits brought by different AI techniques for different sub-tasks within the overall development of explainable decision support tools.

Katie Atkinson
Katie Atkinson
Biography

Katie is Professor of Computer Science, Associate Pro-Vice-Chancellor and Director of the Interdisciplinary Centre for Sustainability Research at the University of Liverpool. She has been conducting foundational and interdisciplinary research on artificial intelligence for over 20 years, with my key areas of research being in the fields of computational models of argument and AI & Law. As Professor of Computer Science, She has published over two hundred articles in peer-reviewed conference proceedings and journals, and have also applied my work in a variety of industrial projects. Her current research is focussed on explainable AI and interdisciplinary applications of this. She develops explainable AI for modelling legal reasoning and also work on projects to bring symbolic AI to the task of digitally-supported materials discovery. She is a member of the leadership team for the UKRI-funded AIchemy hub, supporting adoption of AI for chemistry across the UK. In 2016-2017 she served as President of the International Association for AI and Law, and since 2020 she has served as a member of the Lawtech UK Panel, a government-backed initiative to help transform the UK legal sector through technology.

Submission Information

We will accept two types of paper prepared using the EPTCS LaTeX style.

Papers should be submitted using OpenReview. The paper type should be included as part of the paper title. Reviews will be single-blind. Please note that OpenReview can take up to two weeks to approve a new profile registration that does not contain an institution email address.

We intend to host informal workshop proceedings on OpenReview.

Important Dates

Program Chair

Louise Dennis, University of Manchester
Marija Slavkovik, University of Bergen
Raynaldio Limarga, University of Manchester
Sarah Moth-Lund Christensen, University of Sheffield
Tuva Bardal, University of Warwick
Marina De Vos, University of Bath

Program Committee