ARTIFICIAL INTELLIGENCE FOR DETECTING PUBLIC PROCUREMENT FRAUD THROUGH ESG RISK INDICATORS: A QUALITATIVE SECONDARY STUDY OF SDG 16

Abstract

Public procurement absorbs 12.7% of GDP across OECD countries (OECD, 2025), and bribery to win public contracts featured in 57% of concluded foreign-bribery cases in the OECD’s 2014 analysis. This study asks whether AI-enabled screening that incorporates environmental, social and governance (ESG) indicators can deliver accountable procurement-integrity gains, and under what conditions. It uses a qualitative, theory-informed secondary comparative design: three core cases (Brazil, Ukraine, Italy), two supplier-side illustrations (Odebrecht, Grenfell Tower) and two algorithmic-accountability comparators (the Netherlands, Australia), assessed against a five-level evidence hierarchy running from operational scale to verified outcomes. Governance and procedural indicators are operational; no reviewed case demonstrates validated environmental or social integration. Reported benefits mostly measure exposure, activity or self-reported savings; low selectivity, signal decay and weak follow-through limit effect; and automated enforcement without due process has caused severe harm elsewhere. The paper contributes the ESG-Integrity Analytics Framework (EIAF), five testable propositions and a replicable evidence-grading protocol, and concludes that AI’s contribution to SDG 16.5, 16.6 and 16.10 is conditional on validation, institutional follow-through and accountability.