The United Nations University Institute in Macau (UNU Macau), in collaboration with the Asian Development Bank (ADB) and the United Nations Office for Digital and Emerging Technologies (UN ODET), has released a new research report, AI Systems as Digital Public Goods: Evidence and Recommendations from a Multi-Stakeholder Assessment.
The report is officially announced on the Open Source Program Offices for Good Day at the UN Open Source Week 2026, at the session “AI as a Digital Public Good: Insights from ADB’s Research”.
A digital public good is, above all, a commitment to future beneficiaries. We are not there yet, but this report lays out the conditions that would take us there.
Prof. Tshilidzi Marwala, UNU Rector and UN Under-Secretary-General
The report examines what it would take for artificial intelligence systems to qualify credibly as Digital Public Goods (DPGs): open digital solutions that support the Sustainable Development Goals (SDGs), respect privacy and applicable laws, and are designed to do no harm.
At a time when governments and development partners are looking to AI to improve public services and accelerate progress on the SDGs, the report finds that AI cannot be assessed in the same way as conventional open-source software. AI systems depend not only on code, but also on training data, model weights, documentation, compute infrastructure, deployment settings, and ongoing governance. These features raise complex questions about openness, safety, accountability, and equitable access.
We hope this report will inform policy dialogue and support continued research and innovation in leveraging AIDPGs to address complex development challenges.
Stephanie KC Hung, Director General of ADB’s Information Technology Department, and Antonio G. Zaballos, Director of ADB’s Digital Sector Office
Commissioned by ADB and produced by UNU in partnership with UN ODET, the report draws on a structured desk review, key informant interviews, expert consultations, and a global survey. It highlights a core challenge for the emerging field of AI as Digital Public Goods: openness can support reuse, transparency, and innovation, but it does not automatically ensure public value, local relevance, or protection from harm.
Its central message: the principles that underpin Digital Public Goods remain relevant as ever, but their application must adapt to ensure that AI serves the public interest.
Amandeep Singh Gill, UN Under-Secretary-General and Special Envoy for Digital and Emerging Technologies at UN ODET
The report identifies four major findings. First, openness in AI is multi-dimensional, involving code, model weights, training and evaluation data, documentation, licensing, and the timing and conditions under which components are released. Second, openness does not automatically guarantee public benefit or SDG alignment. Third, governance must be treated as a lifecycle process rather than a one-time certification. Fourth, equity depends on enabling conditions, including shared compute, local-language data, local evaluation capacity, sustainable financing, and institutional readiness.
To address the challenges identified, the report proposes ten recommendations organized under the SAFE framework: Standards, Accountability, Finance, and Equity.
- Under Standards, the report recommends adopting the Model Openness Framework as a reference guideline, establishing public-interest data stewardship pathways, building public-sector readiness evidence into AI-DPG assessment, and creating a public-value and risk annex.
- Under Accountability, it calls for stronger governance support around the existing Digital Public Goods Alliance ecosystem and for responsibility maps and redress mechanisms across the AI value chain.
- Under Finance, it recommends public-interest compute access strategies and linking funding, procurement, and renewal decisions to periodic reporting on outcomes.
- Under Equity, it calls for investment in local-language and domain-specific data as public-good infrastructure, as well as stronger local AI evaluation and audit capabilities, particularly in developing-country contexts.
The report contributes to ongoing global discussions on responsible AI, open-source technologies, digital public infrastructure, and implementation of the Global Digital Compact. It provides practical recommendations for governments, funders, multilateral institutions, standard-setting bodies, and implementers working to ensure that AI supports inclusive and sustainable development.
Access the Full Report
AI Systems as Digital Public Goods: Evidence and Recommendations from a Multi-Stakeholder Assessment
