Governed Interdependence as a Framework for Artificial Intelligence Sovereignty in Somalia

Artificial intelligence (AI) sovereignty is often associated with domestically controlled infrastructure, data localisation and nationally developed models. Such measures may be valuable, but they cannot by themselves preserve public authority when consequential functions rely on AI systems that institutions cannot adequately inspect, constrain, challenge, evaluate or replace. This Perspective distinguishes ordinary technological interdependence from algorithmic dependence and proposes governed interdependence as a more practicable sovereignty objective for resource-constrained states. Drawing on scholarship on technological sovereignty and public-sector AI governance, together with documented policy developments in Somalia, we identify five tests of meaningful public control: legibility, enforceability, contestability, reversibility and independent evaluative capability. For Somalia, these tests suggest three priorities: a risk-tiered register of consequential public-sector AI systems, procurement safeguards that secure oversight and exit rights, and independent Somali-language evaluation connected to accessible human redress. These priorities are prospective policy proposals rather than findings from an empirical assessment of existing government AI deployments. The framework reframes AI sovereignty as the demonstrated capacity to govern dependence, providing a basis for comparative policy analysis and future empirical evaluation in low-resource settings.