On regulatory and ethics readiness for data and AI in vaccine research

Vaccine research is now inseparable from the data infrastructures, artificial intelligence (AI) methods, and genomic resources that generate, link, model, and monitor its evidence. This position paper argues that the governance question this raises turns less on any single tool than on the readiness of the institutions asked to review, regulate, and oversee the use of AI in this research. Readiness is a standing property of institutions, built ahead of need, and ethics belongs in the architecture of a research programme rather than at a checkpoint added at the end. The paper sets out what makes vaccine research a distinctive test for data and AI governance, using Crimean-Congo haemorrhagic fever (CCHF) and severe dengue as cases — noting that CCHF is a zoonosis whose governance architecture must reach across the human-animal-environmental interface that One Health frameworks address. It proposes a layered standards architecture connecting international guidance, national counterparts, and local instruments, and introduces the direction of travel in regulatory governance toward reliance and convergence, from the WHO Global Benchmarking Tool and Listed Authority status to the African Vaccine Regulatory Forum's joint reviews, the African Medicines Agency, and the International Conference of Drug Regulatory Authorities, pointing toward a one world, one dossier architecture in which a dossier assessed in one setting need not be rebuilt entirely in the next. Pre-approved protocols extend this logic into ethics review. The paper treats data movement and sovereignty, AI in safety surveillance (naming the WHO Programme for International Drug Monitoring and VigiBase), authorship, and the monitored emergency use of unregistered and investigational interventions (MEURI) as the places where readiness is won or lost. It makes the case for cooperation with the Human Genome Project II (HGP2) on terms under which participating communities take part as parties rather than as sources of data, and closes on one coherent set of ethical commitments carried across research, assessment, and publication. The closing section makes explicit that artificial intelligence has not created the difficulties the paper describes: it has made visible what was already the case, that ethics committees and regulatory systems were designed for a different scale and a different kind of research, and that the question of what governance for this research requires is now open to the researchers who understand the science best.