An online side event to the UN Global Dialogue on AI Governance.
Monday, 6 July 2026, 07:30–08:30 CEST (05:30-06:30 UTC)
South Asia wants to build AI — but who controls the infrastructure it runs on?
Join this event, led by CODATA Connect, to deliberate on one of the most consequential and least resolved questions in global AI governance: the gap between sovereign ambition and ground reality, and what it means for the communities most at risk.
Registration: https://us02web.zoom.us/meeting/register/9XONUiOGTaiR7iwbv3eObw
South Asia’s AI ambitions are clear. Across the region, governments are investing in AI
infrastructure, data platforms, and national strategies with the explicit goal of becoming builders, not just consumers, of AI. Yet a structural tension sits unaddressed at the centre of this agenda. At critical layers of the AI ecosystem, the infrastructure that powers these ambitions is operated outside domestic jurisdiction, on terms governments did not set. The question this panel asks is simple and unresolved: is South Asia building sovereign AI capacity, or deepening dependency in a more sophisticated form?
The answer looks different depending on where you stand. India has invested significantly in AI infrastructure and policy, yet structural dependencies persist. For countries with fewer resources, the gap between ambition and ground reality is wider. For indigenous and tribal communities across the region, the stakes are more immediate still. Their ecological knowledge, territorial records, and oral traditions are entering AI systems without consent, across jurisdictions where no single actor can be held accountable. Legal obligations under ILO Convention 169 and the UN Declaration on the Rights of Indigenous Peoples exist. The current architecture makes them difficult to enforce.
This panel brings together academics from India, Sri Lanka, and Nepal to examine four questions: what is the current state of AI development in their countries; where dependency constrains sovereign ambition; what governance conditions must exist before community data enters AI pipelines; what the international community must do differently to support sovereignty, not just skills.