The real test ahead: scaling AI in government while building trust
Written by Zoi Roupakia
The OECD’s new report, Governing with AI, could not be more timely. It analyses 200 government AI use cases across 11 core public functions - from justice and procurement to tax administration and policy evaluation. While experimentation is widespread, most initiatives remain at the pilot stage. Scaling and documentation are limited, with many governments constrained by gaps in skills, infrastructure, and outdated systems. In 2023, only 15% of governments had an AI investment framework in place.
The forthcoming policy brief on AI governance in Southeast Asia (Cambridge Industrial Innovation Policy, in collaboration with UNIDO and funded by Lloyd’s Register Foundation) highlights the other side of the equation: citizen trust. Countries with stronger AI readiness - supported by robust infrastructure and governance frameworks - report higher public trust in AI. Globally, higher trust in government also correlates with more optimistic views of AI.
Taken together, the findings point to two critical gaps facing governments today:
Internally: limited investment and infrastructure hinder the capacity to scale AI effectively.
Externally: public trust depends on transparent, accountable governance that makes AI use credible.
Addressing both means building the capacity to scale AI while maintaining public trust - the foundation for sustainable adoption.
According to the OECD, unlocking trustworthy AI in the public sector depends on three pillars:
Enablers: investment in data, skills, and infrastructure.
Guardrails: robust rules, transparency, and oversight across the AI lifecycle.
Engagement: structured participation of citizens and civil society in AI policymaking and implementation.
Global frameworks are converging—from the OECD AI Principles to UNESCO’s Recommendation on the Ethics of AI and the Council of Europe AI Treaty—yet national implementation still lags behind rhetoric.
The question is no longer whether governments will adopt AI—they already have. The defining challenge now is which governments can move beyond pilots to systematic transformation, while safeguarding the democratic legitimacy essential for sustainable innovation.
And as a Greek, I found it notable that the OECD also highlights Greece’s DidaktorikaAI, an AI-powered digital library providing access to over 50,000 doctoral dissertations—a reminder that even targeted initiatives can expand access to knowledge and give smaller states visibility in the global AI story.
Sources: Organisation for Economic Co-operation and Development (OECD) (2025), Governing with AI: Enabling trustworthy use of artificial intelligence in the public sector; Cambridge Industrial Innovation Policy (2025), Policymaking for a more resilient world, Executive summaries