Abstract
Emerging zoonoses, particularly vector-borne ones, are driven by climate change, biodiversity loss and land-use change. Surveillance, however, remains fragmented and largely reactive, and often fails to detect the risk signals that precede spillover. This narrative review examines how artificial intelligence (AI) and machine learning (ML), integrated with georeferenced and climatic data, could support a transition toward predictive, anticipatory One Health surveillance. Using tick-borne encephalitis (TBE) and Crimean-Congo haemorrhagic fever (CCHF) as case studies, we distinguish evidence from conventional statistical models, ML models and deep-learning applica-tions. ML models combining climatic, land-cover and host variables have shown promising ret-rospective performance in mapping tick habitat suitability and TBE and CCHF risk, whereas deep learning has been applied mainly to tick identification. However, AI evidence specific to TBE and CCHF remains limited: few disease-specific studies exist, validation is largely internal, no model has been evaluated prospectively, and some reported figures derive from synthetic data. We consider this gap a key finding. We also discuss data quality, model transferability, validation practices, al-gorithmic transparency and the EU AI Act. AI may enable a shift from costly emergency response toward validated Early Warning Systems if methodological, infrastructural and regulatory gaps are addressed. We outline priorities for researchers, veterinarians, clinicians and policymakers.