EXCALIBUR Research Presented at IFCS 2026

EXCALIBUR was represented at IFCS 2026, the International Federation of Classification Societies Conference, held in Milan, Italy.

Sofia Vei, PhD candidate at Aristotle University of Thessaloniki and EXCALIBUR researcher at the Data and Web Science Lab, presented the conference contribution “Not All Harms Hurt Equally: Understanding AI Safety Through Ordinal Severity.”

The work was co-authored by Sofia Vei, Paolo Giudici, Pavlos Sermpezis, Athena Vakali, and Adelaide Emma Bernardelli.

The presentation discussed how ordinal severity can help support more meaningful approaches to AI safety, especially when AI harms affect different stakeholder groups in different ways. Rather than treating all harms as equally severe, the work highlights the importance of understanding which impacts require greater attention and prioritization.

The discussion following the talk was particularly valuable, with thoughtful questions and interest in how ordinal perspectives can contribute to clearer, more human-aware AI risk assessment.

Presented by: Sofia Vei
Conference: IFCS 2026, International Federation of Classification Societies Conference
Location: Milan, Italy
Contribution title: “Not All Harms Hurt Equally: Understanding AI Safety Through Ordinal Severity”