What happened
Published 7 May 2026 as OECD Artificial Intelligence Papers No. 58, this 60-page working paper draws on consultations with 15 public audit institutions across 14 countries and the European Union (including the UK NAO, Singapore AGO, Netherlands Court of Audit, European Court of Auditors, and others) to map where supreme audit institutions and public audit bodies actually stand on AI adoption. The headline finding is stark: 'AI adoption in public audit remains at an early stage' — the gap between pilot projects and scalable operational deployment is wide. Emerging AI applications include anomaly detection, document processing, knowledge management, and predictive risk assessment. The paper identifies three structural bottlenecks preventing institutions from scaling: fragmented data systems, limited internal technical expertise, and evolving governance frameworks. It argues that strengthening data governance, digital infrastructure, and internal development capacity will be critical for audit institutions to responsibly scale AI while maintaining the transparency and accountability that public trust demands.
Why it matters
For governments and regulators, this is the clearest cross-country baseline available on AI in oversight functions — it directly informs decisions about resourcing, upskilling, and data infrastructure at audit institutions; for private sector firms, it signals that their principal public oversight bodies are at an early stage of AI adoption, which has implications for the near-term credibility and speed of AI-assisted regulatory examination.
Action needed
Government CIOs and CAOs should benchmark their audit institution's current AI maturity against the three structural bottlenecks identified, and use the report as input to digital transformation roadmap prioritisation for the next budget cycle.