Risks connected with AI systems are now a recurrent topic in public and academic debates. In this talk I will offer a philosophical analysis of such risks with the aim of providing a comprehensive and general framework to think about AI-related risks. First, by building upon the existing literature on natural disaster risk analysis, I will present a multi-component analysis of risk and discuss its benefits in terms not only of conceptual clarification but also of possible mitigation. Second, by adopting a multi-hazard and multi-risk framework, I will analyze AI-related risks as potentially emerging from the combination of the risks related to the different elements of the AI system and its context of use.
Official event link: https://caiml.org/dighum/announcements/viola-schiaffonati-ai-related-risks-a-philosophical-analysis-2026-05-05/