Artificial Intelligence and Corruption/Fraud Detection in Nigeria's Public Sector

Background to the Study

Corruption—abuse of entrusted public power for private gain—undermines service delivery and trust. Nigeria’s CPI 26/100 in 2025, shared with Cameroon, Guinea, Kyrgyzstan, Guatemala, Papua New Guinea, unchanged from 2024 while rank fell from 140 to 142, signals stagnation despite reforms. Traditional methods—manual audit, whistleblower, complaint-driven investigation—are inherently limited in speed and scale.

AI offers pattern-recognition, anomaly detection, and machine learning on large transactional and administrative datasets. In Nigerian financial services, research finds NLP-augmented AI significantly moderates fraud detection accuracy, yet 2025 banking study finds adoption slow, fragmented due to costs and limited expertise. What is less examined is AI within public anti-corruption and revenue agencies—EFCC, ICPC, FIRS—as distinct from commercial banking. The UK’s International Public Sector Fraud Forum (IPSFF), multinational body sharing best practice, published framework describing AI as “next great technological frontier” for public-sector counter-fraud, yet no comparable Nigerian framework or systematic empirical assessment exists for federal anti-corruption institutions. This gap motivates this study. Transparency International Corruption Perceptions Index 2024 and UK IPSFF guidance on counter fraud | Related AI and accounting information systems project topics

Statement of the Problem

Nigeria’s stagnant CPI performance raises question whether anti-corruption practice, including technological dimension, is adequate. While AI fraud detection researched rigorously in banking sector examining cost, compliance, competency barriers, marked absence exists of equivalent empirical research within public-sector anti-corruption and revenue agencies—EFCC, ICPC, FIRS—on their own AI/analytics detection capability. Without evidence on state of AI detection within own anti-corruption institutions, difficult to assess whether poor CPI reflects downstream resourcing/political-will/enforcement problem or upstream detection-capability problem or both—distinction with direct implications for investment priorities. Study addresses gap by examining extent of AI adoption, institutional and data-readiness factors shaping adoption, and relationship between adoption and perceived effectiveness.

Aim and Objectives

Aim: Examine extent and perceived effectiveness of AI-driven fraud and corruption detection within selected federal anti-corruption and revenue agencies.

Objectives:
1. Assess current level of AI/analytics adoption for fraud and corruption detection.
2. Identify institutional and data-readiness factors facilitating/hindering adoption.
3. Evaluate staff-perceived effectiveness of AI tools in detecting fraudulent/corrupt activity.
4. Determine relationship between adoption and perceived effectiveness.
5. Recommend measures for strengthening AI-enabled detection in Nigerian public service.

Research Questions

1. What is current level of AI/analytics adoption for detection in selected federal agencies?
2. What institutional and data-readiness factors facilitate/hinder adoption?
3. How effective do staff perceive AI tools to be?
4. What is relationship between adoption and perceived effectiveness?
5. What measures can strengthen AI-enabled detection?

Research Hypotheses

H₀₁: No significant relationship between AI adoption and perceived effectiveness.
H₀₂: Institutional and data-readiness factors (data quality/availability, analytical skills, leadership commitment) have no significant influence on adoption.
H₀₃: No significant difference in adoption across selected agencies.

Significance of the Study

Contributes to public administration scholarship by extending AI-fraud-detection agenda from banking to public anti-corruption institutions. Theoretically applies Fraud Triangle, Fraud Diamond, Principal-Agent theories to Nigerian agency context. Practically valuable to EFCC, ICPC, FIRS for maturity benchmarking; to NFIU and CBN for cross-institutional data-sharing coordination; to National Assembly for assessing whether CPI stagnation reflects detection capability gap warranting investment; to academics for validated instrument measuring AI adoption maturity in anti-corruption contexts. For citizens, more effective AI detection offers pathway to reduced corruption and improved trust. See public sector reform and anti-corruption research guides

Scope of the Study

Delimited to selected federal anti-corruption and revenue agencies in Abuja FCT. Focuses on AI/analytics-driven detection—pattern-recognition, anomaly detection, ML tools applied to investigative, audit, revenue data—as distinct from general AI in public service delivery. Empirical focus on perceptions and institutional knowledge of investigative, audit, ICT/data-analytics staff via cross-sectional questionnaire, complemented by document analysis of agency reports and international frameworks.

Limitations

Cross-sectional design limits causal inference between adoption and effectiveness. Sensitive/classified nature of investigative info restricts questionnaire to general practice/technology items, not case-specific detail, limiting granularity vs internal audit. Self-reported effectiveness may not align with independently verified case outcomes outside research access. Sample restricted to Abuja federal agencies limits generalizability to state-level bodies or agencies with different mandates.

Operational Definition of Terms

Artificial Intelligence / Analytics-Driven Detection: Use of machine-learning, pattern-recognition, advanced analytical software to identify anomalous patterns indicative of fraud/corrupt practice.

Fraud: Intentional deception for unlawful gain, encompassing asset misappropriation, financial-statement fraud, corruption.

Corruption: Abuse of entrusted public power/position for private gain.

Detection Effectiveness: Extent to which tool/system succeeds in identifying fraudulent/corrupt activity.

Data Readiness: Extent to which organization data—quality, accessibility, structure—is suitable for AI/analytics detection.

Anti-Corruption Agency: Government institution with statutory mandate to investigate, prevent, prosecute corruption, e.g., EFCC or ICPC.

EFCC Nigeria official publications | ICPC Nigeria mandate and reports | U.S. GAO AI accountability framework for public sector

Conclusion

Nigeria’s CPI 26/100 rank 142/182 reflects stagnation despite EFCC/ICPC existence. AI offers frontier for scaling detection beyond manual audit, as recognized by UK IPSFF. While banking sector research advances, public anti-corruption agencies remain empirically underexamined regarding AI adoption. This study’s survey of Abuja federal agencies using Taro Yamane sampling and Likert constructs aims to establish adoption level, identify data quality, skills, leadership factors shaping adoption, and test relationship with perceived effectiveness via Chi-square, regression, ANOVA. Expected outcome: baseline maturity assessment informing recommendations on data infrastructure standardization, analytical capacity building, inter-agency data sharing agreements, and governance frameworks for ethical AI use. For implementation models, see our AI in governance project materials collection

FAQs

1. What is Nigeria's corruption ranking in 2025?

Transparency International CPI 2025 scored Nigeria 26/100 and ranked 142nd of 182 countries, stagnant from 2024, behind 33 other African countries.

2. Why is AI relevant for corruption detection?

AI can detect anomalous transactional and administrative patterns at scale and speed unattainable through manual audit and reactive investigation.

3. Which agencies does this study focus on?

Selected federal anti-corruption and revenue agencies in Abuja FCT, including EFCC, ICPC, and FIRS, focusing on investigative, audit, and ICT staff.

4. What theories underpin the study?

Fraud Triangle Theory, Fraud Diamond framework, and Principal-Agent Theory applied to anti-corruption institutional context.

5. What is the IPSFF AI framework?

UK International Public Sector Fraud Forum guidance describing elements public agencies should consider when deploying AI against fraud, calling AI the next great technological frontier.

6. What are institutional barriers to AI adoption in public sector?

Data quality/availability gaps, limited analytical skills, high implementation costs, fragmented data sharing, and leadership commitment deficits.

7. How is AI adoption measured in the study?

Structured 5-point Likert questionnaire measuring three constructs: adoption level, institutional/data-readiness, perceived detection effectiveness, validated and analyzed via SPSS.

8. Has AI been studied in Nigeria's banking sector?

Yes, extensively. Research shows AI improves fraud detection accuracy but adoption remains slow due to cost and expertise constraints—similar challenges likely in public sector.

9. What are recommendations for strengthening AI anti-corruption capability?

Standardize data infrastructure, build analytical capacity, establish inter-agency data sharing, and develop ethical AI governance frameworks tailored to Nigerian public service.

10. What are limitations of self-reported effectiveness?

Staff perceptions may not align with independently verified case outcomes, and sensitive classified information restricts granularity versus internal agency audit.