UN Study: Most AI Systems Fail Digital Public Good Standards

A newly released 66-page assessment commissioned by the Asian Development Bank and produced by the United Nations University in partnership with the UN Office of Digital and Emergent Technologies examines why so few AI systems qualify as Digital Public Goods (DPGs). The authors conducted a structured desk review of policy and technical frameworks, interviewed cross-sector experts from the UN, governments, civil society, academia, and private sector, and ran a global survey to validate findings across regions and AI readiness levels. The paper argues that while AI is frequently positioned as a DPG for sustainable development, the current DPG Standard is rarely met in practice due to issues around openness, safety, governance, and maintainability.

The report provides specific recommendations for policymakers and developers to close the gap between aspiration and reality. Key recommendations include clearer definitions of openness for AI, stronger safety and accountability mechanisms, and more inclusive governance structures. The study highlights that without these changes, the term 'AI as Digital Public Good' risks becoming an empty label. The assessment is part of a broader push to align AI development with the Sustainable Development Goals and the Global Digital Compact's call for safe, open-source software and AI models that benefit society as a whole.