AI for All or AI for Some? Governing Artificial Intelligence for Educational Sovereignty in the Majority World

Abstract SOAR artificial intelligence is rapidly reshaping education. Adaptive tutors, biometric attendance, and predictive dashboards are no longer prototypes; they are embedded in classrooms and ministries. For the Majority World, this acceleration brings a paradox: AI can expand access, support multilingual learning, and ease teacher shortages, but it can also tighten dependence on infrastructures and epistemologies rooted in the Minority World. The central question, as sovereignty scholars from Puiggrós to Hajji warn, is not whether AI works but whether it strengthens or erodes educational sovereignty. This article develops the SOAR™ governance framework—sovereignty, ownership, accountability, resilience—as both lens and practice. Drawing on decolonial pedagogy, digital colonialism, and comparative governance, it treats sovereignty as the hinge between aspiration and implementation. The method combines a critical review of the literature with 11 comparative case studies: Brazil, India, China, the United Arab Emirates, Kenya, Uruguay, Rwanda, Colombia, Ghana, Malawi, and Hawai‘i. Findings show sovereignty is contested, partial, and sometimes symbolic. China secures infrastructure but limits plural ownership; Uruguay empowers teachers but struggles with resilience; Kenya and Malawi reveal donor dependence that corrodes every pillar of SOAR; and Hawai‘i demonstrates epistemic sovereignty even under infrastructural constraint. Placed against UNESCO’s ethical baselines, the OECD’s innovation agenda, and the World Bank’s market-first approach, these cases reveal what global frameworks omit: sovereignty. Without it, AI in education will remain “AI for some.” With it, “AI for all” moves from aspiration to possibility.