New Publication in Artificial Intelligence: AI Harmonics

A new journal article, AI Harmonics: A human-centric and harms severity-adaptive AI risk assessment framework, has been published in Artificial Intelligence.
The paper introduces AI Harmonics (AIH), a stakeholder-oriented approach for ranking and comparing AI harms when exact numerical severity scores are unavailable or unreliable. Instead of assigning arbitrary numbers to severity judgments, AIH uses ordinal information to examine where severe harms are most concentrated across affected stakeholder groups.

By applying the framework to AI incident data, the study shows how harm prioritization can move beyond abstract scoring and focus more directly on those who are most affected. This direction is closely aligned with EXCALIBUR’s focus on trustworthy, responsible, and human-centric AI assessment.

Authors: Sofia Vei, Paolo Giudici, Pavlos Sermpezis, Athena Vakali, Adelaide Emma Bernardelli
Journal: Artificial Intelligence
DOI: 10.1016/j.artint.2026.104587