Countries around the world are regulating artificial intelligence in wildly different ways, creating a patchwork of conflicting rules that makes it harder to keep AI systems safe and trustworthy. A new policy report from the United Nations University Institute in Macau reveals the core problem: without better coordination between nations' AI safety frameworks, the world risks both inadequate protection and stifled innovation. The report, which studied governance approaches in China, South Korea, Singapore, and the United Kingdom, offers concrete recommendations for building a more unified global approach to AI safety .

Why Does AI Governance Fragmentation Matter?

Imagine trying to drive across four countries where each one has completely different traffic laws, speed limits, and vehicle safety standards. That's essentially what AI developers and deployers face today. When regulations diverge sharply, companies struggle to build systems that comply with multiple jurisdictions simultaneously, and safety risks can slip through the cracks when rules don't align. The UN University report emphasizes that interoperability, a cornerstone of effective AI governance, is essential for reducing risks, fostering innovation, enhancing competitiveness, promoting standardization, and building public trust .

The challenge is significant. Progress toward coordinated AI governance remains hindered by fragmented regulations, limited global coordination, and insufficient engagement from the Global South, which represents billions of people whose voices are largely absent from the conversation .

What Are the Three Pillars of Better AI Governance?

The UN University report identifies three critical dimensions where countries need to align their approaches to AI safety. These dimensions span ethical, regulatory, and technical domains, each addressing a different layer of how AI systems are developed, deployed, and overseen. The researchers analyzed how countries handle autonomous vehicles, education applications, and cross-border data flows to understand where alignment is possible and where fundamental differences exist.

• Ethical Interoperability: Advancing a global AI ethics framework that reflects shared values across cultures and legal systems, ensuring that fundamental principles like fairness and transparency are understood consistently.
• Regulatory Interoperability: Establishing multilateral systems for coordinated governance and enhancing transparency and accountability mechanisms so that rules in one country don't contradict those in another.
• Technical Interoperability: Promoting interoperable digital public infrastructure and embedding interoperability by design in technical standards, so that AI systems can operate across borders without requiring complete redesigns.

The report's approach is notably practical. Rather than imposing a one-size-fits-all solution, it adopts what researchers call a "regulatory learning" approach, engaging stakeholders in each jurisdiction to capture local insights and foster learning between countries . This collaborative model recognizes that different nations have legitimate reasons for their regulatory choices, rooted in their legal traditions, cultural values, and economic priorities.

How Can Policymakers Build Better AI Safety Frameworks?

The UN University report offers specific, actionable recommendations for governments and international bodies looking to strengthen AI governance while maintaining flexibility for local contexts. These steps move beyond abstract principles toward concrete mechanisms that can actually be implemented.

• Deepen Normative Specificity: Move beyond vague principles like "fairness" and "transparency" to define precisely what these terms mean in practice, so that regulators and companies share a common understanding of what compliance requires.
• Enhance Legal Interoperability: Work toward mutual recognition agreements and harmonized definitions across jurisdictions, reducing the burden on companies that must navigate multiple regulatory regimes simultaneously.
• Expand Interoperable Standards: Develop technical standards that allow AI systems to function across borders without major modifications, similar to how electrical standards or telecommunications protocols work globally.
• Strengthen Data Governance: Create frameworks for secure cross-border data flows that respect privacy and security concerns while enabling the data sharing necessary for AI development and deployment.
• Invest in AI Literacy and Capacity Building: Ensure that policymakers, regulators, and civil society in all regions, especially the Global South, have the knowledge and resources to participate meaningfully in AI governance decisions.

The report concludes that the future of AI safety governance is evolving toward evidence-based, outcomes-oriented models that complement principle-led frameworks . This shift reflects a maturation in how the world thinks about AI regulation, moving away from purely aspirational statements toward measurable results and accountability.

What Makes This Report Different From Other AI Governance Discussions?

Most AI governance conversations focus on what individual countries or regions should do. This UN University report takes a different angle by examining how countries can work together more effectively. The study was a joint effort by multiple UNU Global AI Network members, including researchers Yik Chan Chin, David A Raho, Hag-Min Kim, Chunli Bi, James Ong, Jingbo Huang, and Serge Stinckwich, and was sponsored by SenseTime .

By comparing frameworks across four diverse jurisdictions, the researchers identified both areas of convergence, where countries' approaches naturally align, and areas of divergence, where fundamental differences require creative solutions. This comparative approach provides a roadmap for how other countries can learn from these examples and contribute to building a more coherent global AI governance ecosystem.

The stakes are high. As AI systems become increasingly powerful and integrated into critical sectors like healthcare, education, and infrastructure, the lack of coordinated safety standards poses real risks. At the same time, overly fragmented or contradictory regulations can slow beneficial innovation and create unfair competitive advantages for companies in less-regulated jurisdictions. The UN University report suggests that the path forward requires sustained commitment from policymakers to deepen normative specificity, enhance legal interoperability, expand interoperable standards, strengthen data governance, and invest in AI literacy and capacity building across all regions .