The Global AI Governance Gap: Why Policies Are Failing to Protect Citizens
As artificial intelligence rapidly embeds itself into public services, policing, and healthcare, a sobering reality is emerging: governments are moving too slowly to protect their citizens. According to the newly released second edition of the Global Index on Responsible AI (GIRAI), the acceleration of AI investment has vastly outpaced the creation of enforceable, rights-respecting governance. While national AI strategies and voluntary frameworks are proliferating, the institutions and enforcement mechanisms required to prevent algorithmic harm are lagging dangerously behind.
Produced by the South Africa-based Global Center on AI Governance, the 2026 index evaluates 135 countries based on their commitment to human rights, ethics, and inclusion. The findings reveal a stark global divide. Although 53 percent of the world’s population has used generative AI, the average global governance score sits at a meager 35 out of 100. More concerning is the gap between policy and practice: evidence of actual implementation exists in only 55 percent of cases where frameworks are active.
“Responsible AI cannot be secured through principles alone,” stated Rachel Adams, founder and CEO at the Global Center on AI Governance. “As AI becomes a structural force in public life, governments need enforceable obligations, independent oversight, public disclosure, monitoring systems and accessible routes for redress.”
The Double Standard in Transparency
One of the report’s most glaring findings is the hypocrisy in how governments manage AI transparency. While state regulators increasingly demand accountability from private tech developers, they remain opaque about their own use of automated systems. The weakest-performing area globally is AI use in public service delivery—systems that dictate access to welfare, housing, and migration. Only 18 percent of surveyed countries require the public disclosure of government algorithmic systems. Furthermore, while governments are investing heavily in AI skills for the future economy, they are largely neglecting the immediate rights of workers to organize and collectively bargain in response to AI-driven workplace disruptions.
Africa’s Struggle with Implementation
For Africa, the report presents a complex picture of growing momentum undermined by a lack of enforceability. The continent recorded the world’s lowest average GIRAI score (22 out of 100). Out of 39 African nations surveyed, only six—Nigeria, Egypt, Kenya, Ghana, Benin, and Morocco—scored above the global average.
The primary barrier across the continent is not a lack of ambition, but a heavy reliance on non-binding “soft law.” Only 21 percent of Africa’s documented AI policy frameworks are legally enforceable. However, there is a silver lining: African nations demonstrate strong execution when it comes to labor and skills, recording a 65.5 percent implementation rate in this dimension, which exceeds the global average. Yet, as recent cases of unacceptable-risk AI deployment in Kenya, Ghana, and Uganda highlight, the urgent need for oversight mechanisms cannot be overstated.
As global tech integration accelerates, the overarching warning from the index is clear. Without a shared, rights-based foundation of enforceable rules, the global AI landscape risks prioritizing market interoperability over the protection of vulnerable populations, leaving the Global South to bear the brunt of the consequences.
