Scaling artificial intelligence (AI) in Public Audit Institutions: Bridging the Gap Between Pilots and Practice in Supreme Audit Institutions

Drawing on a systematic review of 60 peer-reviewed publications, international audit standards documents, Organisation for Economic Co-operation and Development (Organisation for Economic Co-operation and Development (OECD)) and International Organisation of Supreme Audit Institutions (INTOSAI), reports, and country-level Supreme Audit Institutions (SAI) innovation studies spanning 2020 to 2026, this paper investigates the structural barriers preventing AI scaling in public audit institutions and develops an integrated responsible scaling framework for Supreme Audit Institutions (SAIs): the Responsible AI Scaling Protocol (Responsible AI Scaling Protocol for Supreme Audit Institutions (RASP-SAI)). Three analytical tables present a comparative assessment of AI adoption profiles across seven leading global SAIs; a structured taxonomy of eight structural barriers with severity ratings and mitigation pathways; and a comparative evaluation of six AI governance and scaling frameworks applicable to SAI contexts. The paper develops an original analytical argument, the Public Audit AI Paradox, that the institutional characteristics that make SAIs effective guardians of public accountability (independence, conservatism, transparency obligations, parliamentary accountability) are simultaneously the characteristics that most obstruct AI scaling, creating a governance dilemma that requires purpose-designed solutions rather than adaptations of private sector AI adoption models. Recommendations are directed at SAI leaders, legislative oversight bodies, the INTOSAI, and international development partners.