Artificial Intelligence (AI) and the New Colonialism of Climate Data in the Global South

Toward just and decolonial adaptation futures Johannes Bhanye Abstract As artificial intelligence (AI) becomes increasingly central to climate adaptation governance, a troubling paradox emerges: the very tools intended to enhance resilience may also deepen global inequalities. AI-driven climate data systems, ranging from predictive flood models and vulnerability indices to remote sensing platforms, are largely designed, funded and governed by institutions in the global North. These systems frequently impose standardised, top-down representations of risk that marginalise the lived knowledge, cultural practices and contextual needs of communities in the global South. This working paper argues that such developments reflect a new form of epistemic colonialism, whereby the power to define climate vulnerability and adaptation priorities is abstracted from the people most affected. Without mechanisms for local participation, ethical oversight and knowledge pluralism, ‘smart’ adaptation threatens to replicate historical patterns of exclusion, under the guise of innovation. To ensure that AI serves as a tool for justice, not dispossession, we call for a paradigm shift in climate governance. Key recommendations include: (1) strengthening data sovereignty by investing in local AI infrastructure and training; (2) enabling participatory AI design grounded in community knowledge systems; and (3) promoting epistemic pluralism that values indigenous, informal and experiential knowledges as critical to climate resilience.