Turning measurement into momentum so agile governance can keep pace with AI

Good regulation starts with good information but, in artificial intelligence (AI), it's rare. Data on how AI systems are built, trained and deployed remains fragmented — scattered across private companies, buried in proprietary processes or simply never collected at all.

This means governments trying to regulate AI often have to work with incomplete snapshots that use voluntary disclosures and inconsistent reporting standards, which provide limited visibility into what's actually happening inside these systems. The result is that regulators are asked to manage risks like algorithmic bias and data misuse without the evidence they need to understand them, let alone to address them effectively or consistently across borders.

Various indices and studies have emerged in recent years to help governments and industry assess where AI governance stands. These include the OECD’s AI Policy Observatory, the AI Readiness Index by Oxford Insights, Stanford University’s AI Index Report and the Global AI Index from Tortoise Media. Each contributes valuable insights, but differs in focus. Some emphasize innovation and investment, while others assess ethics, governance or human capital readiness.

The AI Governance International Evaluation (AGILE) Index is the latest addition to this landscape. Covering more than 40 countries, it offers a comparative snapshot of how legal frameworks, institutional capacity and societal safeguards are evolving – and where they’re not.

The real value of an index like this isn't in ranking countries, but in revealing the gaps and opportunities in AI development. Some countries have strong strategy documents but weak enforcement. Others have clear mandates but fragmented oversight. In many countries, innovation policies are moving faster than legal or ethical guardrails are being established.

These differences matter. Without a shared evidence base, AI regulation risks becoming reactive or symbolic, rather than enabling and effective.

Regulatory aspiration often develops faster than the institutions designed to uphold it and reaches varying degrees of operational maturity. High-income countries tend to have stronger regulatory structures and technical capacity, for example. On the other hand, many middle-income countries show public awareness and policy intent but struggle with enforcement and coordination. In some contexts, regulatory conversations have happened, but have not necessarily been translated into mandates, oversight or institutional coordination.

But ambition without infrastructure is just aspiration. AI governance must go beyond principles to cover enforcement, implementation and accountability.

The AGILE Index 2025 was developed by a consortium of research institutions in China to assess national AI governance across 43 legal, institutional and societal indicators. It groups countries into four profiles ranging from those with more mature frameworks, called “all-round leaders”, to those still laying basic foundations, referred to as “foundation seekers”.

The index aims to capture the structural reality of AI policy systems worldwide and the latest version provides three takeaways in this respect:

By laying bare national strengths and blind spots, the AGILE Index encourages governments to move past press releases and into capability-building. The index shows that countries with publicly available AI governance data — such as regulatory frameworks and accountability mechanisms — score markedly higher across institutional maturity indicators. Transparency itself becomes a driver of progress, enabling peer benchmarking and policy learning.

There are deep asymmetries in institutional preparedness for AI between regions, income groups and even within governments. Some ministries of innovation surge ahead while their country's legal bodies lag behind. Elsewhere, policymakers hesitate to act on AI because they lack the tested frameworks to do so.

AGILE’s institutional dimension highlights a gap of more than 40 percentage points between high- and middle-income countries in regulatory implementation capacity. Even within advanced economies, oversight and data protection bodies often trail innovation agencies, revealing the internal fragmentation the index quantifies.

Several countries, including Singapore, the UK and South Korea, have demonstrated that agile governance is not just theoretical, it's operational. But that progress must be shared, studied and scaled.

These countries are top performers in the AGILE Index because they have embedded agile mechanisms — from regulatory sandboxes to ethics-by-design frameworks — that translate principles into practice. Their experience offers a template for others to build flexible but accountable AI systems.

Governance must evolve as rapidly as the innovations it seeks to guide. Launched in July 2025, the Global Regulatory Innovation Platform (GRIP) aims to encourage this by bringing together public and private stakeholders to learn from best practices around the world, co-develop practical frameworks and build communities of practice around agile, inclusive governance.

Initiatives like the AGILE Index provide the dashboard for this, but GRIP provides the steering wheel. While indices can offer measurement, GRIP enables movement, building the connective tissue between data, policy and real-world reform. Whether it's through policy roundtables, shared toolkits or national self-assessment pilots, governments must move faster from principles to prototypes, and from working papers to working policies.

Measuring maturity is important; building it is essential

This means clarifying the distance between ambition and capability. Trust in AI must be secured through principles but also through enforceable systems that give businesses and citizens clarity and confidence. Countries that are willing to measure their progress in these areas are already ahead of the curve.

For those ready to act, the door is open. In the intelligent age, AI governance that stands still is already falling behind.