Researchers involved in the DECIDE project have presented a new conference paper entitled “News-Informed Probabilistic Models for AI Risk Analysis” at CAiSE 2026.
The paper introduces a novel approach for assessing AI-related risks by combining probabilistic modelling, AI governance frameworks, and real-world incident reports extracted from news sources.
The proposed method supports the development of practical tools for AI risk assessment and contributes to increasing transparency and risk awareness in the use of AI technologies.
The research builds on concepts from the EU AI Act and the IPCC risk framework, proposing a data-driven methodology for identifying hazards, vulnerabilities, and exposure factors associated with AI systems.
The work was developed by Mattia Fumagalli, Stefano M. Nicoletti, Diego Calvanese, and Giancarlo Guizzardi within the broader context of research on trustworthy and transparent AI systems.
The full conference paper is available here.