Why AI Governance Frameworks Keep Treating Sovereignty as a Technical Parameter

When a regulatory sandbox for AI is launched, it tests technical safety, bias metrics, and robustness benchmarks. The participants are technology companies, the regulators are technology regulators, the civil society observers comment on technical reports. The governance question — who decides what safety means for this society — is never asked. The sandbox is technical. Sovereignty is absent.The dominant paradigm in AI governance is technical risk management. Identify the hazard, measure the probability, mitigate the impact, monitor the residual. This paradigm works for engineering. It fails for sovereignty.