One of the world's oldest universities, home to the Center for Reasoning, Normativity and AI (CERNAI), which conducts AI safety and alignment research led by Prof. Federico Faroldi.
One of the world's oldest universities, home to the Center for Reasoning, Normativity and AI (CERNAI), which conducts AI safety and alignment research led by Prof. Federico Faroldi.
People
Updated 05/18/26Professor of Ethics, Law and Artificial Intelligence; Director of CERNAI and Normative Risk Lab
Funding Details
Updated 05/18/26- Annual Budget
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26The University of Pavia (Università degli Studi di Pavia) is one of the oldest universities in the world, with origins tracing to an imperial rhetoric school established in 825 and the formal founding of its Studium Generale by Charles IV in 1361. Located in Pavia, Lombardy, approximately 35 km from Milan, the university currently enrolls about 26,400 students across 18 departments and 105 degree programs, supported by roughly 1,800 professors, lecturers, and administrative staff. It is internationally recognized as a research university with over 500 international partnerships and a long tradition of scientific distinction, counting among its alumni and faculty Alessandro Volta and Nobel laureates Camillo Golgi, Giulio Natta, and Carlo Rubbia. The university's relevance to AI safety and existential risk reduction is centered on the work of Prof. Federico L.G. Faroldi, who holds a Chair in Ethics, Law, and Artificial Intelligence in the Department of Law. In September 2025, Faroldi launched CERNAI (Center for Reasoning, Normativity and AI), an interdisciplinary research center focusing on logic, normativity, and AI systems. CERNAI is funded by a grant of over €1.2 million from Italy's Ministry of University and Research through the Italian Fund for Science (FIS), awarded for the research project "Generic Reasoning." Faroldi also directs the Normative Risk Lab, an independent research body that examines how norms create or fail to prevent catastrophic and existential risks, with a particular focus on artificial intelligence. Faroldi is an affiliate faculty member at UC Berkeley's Center for Human-Compatible AI (CHAI) and a member of the AI Safety Community Faculty at the Future of Life Institute. His technical research addresses AI alignment by integrating normative reasons into reinforcement learning to address problems such as goal misgeneralization and scheming AI systems. His conceptual and policy work focuses on operationalizing existential risk within established risk management standards (e.g., ISO 31000:2018) that are being incorporated into law, including the EU AI Act. In April 2024, Open Philanthropy recommended a grant of €22,000 to the University of Pavia to support Faroldi's development of a course titled "Law, Ethics, and AI: Alignment and Normative Risk," which covers topics including the alignment and control problems, normative uncertainty, and proposed EU AI regulation.
Theory of Change
Updated 05/18/26The University of Pavia's AI safety work, primarily through Prof. Faroldi and CERNAI, operates on two complementary tracks. On the technical side, integrating formal normative reasoning (deontic logic, reasons-based approaches) into AI training aims to produce more robustly aligned systems that generalize goals correctly and resist scheming behaviors. On the governance side, embedding precise definitions of existential and catastrophic risk into legal and regulatory frameworks—such as the EU AI Act and ISO risk management standards—creates institutional and legal structures that constrain dangerous AI development. Training future researchers and policymakers through courses on AI alignment and normative risk builds the human capital needed to sustain this work over time.
Grants Received
Updated 05/18/26Projects
Updated 05/18/26CERNAI is a research center at the University of Pavia dedicated to the study of reasoning, normativity and artificial intelligence at the intersection of law, philosophy and logic, with research lines on generic reasoning, genericity-aware AI alignment, law-following AI and formal models of agentic and existential risk.
Discussion
No comments yet. Be the first to share your thoughts.