Miguel’s expertise spans frontier AI governance and safety, technical evaluation and assurance, public-sector AI governance, and the institutional response to emerging risks from advanced AI systems. His work focuses particularly on harmful manipulation, information integrity, democratic resilience, systemic risk and the evaluation of model and agentic capabilities. He is interested in how governments can build technical capacity to assess frontier systems, translate uncertain technical evidence into proportionate policy and supervisory decisions, and design institutions capable of anticipating rapidly evolving AI risks. His work also covers AI strategy, risk assessment, evaluation frameworks, governance of general-purpose AI, and the interface between technical research, regulation and public-sector implementation.