Received: October 14, 2025
Accepted: February 11, 2026
Structural corruption, which undermines development and erodes trust in public institutions, can be addressed through the use of Artificial Intelligence (AI)—a transformative tool with the potential to strengthen transparency in the public sector. Against this backdrop, the present study aims to identify and analyze the characteristics of the most significant AI-based practices used worldwide to combat corruption. To this end, a qualitative, descriptive, and comparative approach was adopted, based on the examination of 19 international initiatives. The findings reveal a global trend toward the use of AI—through supervised machine learning—primarily in areas that are highly vulnerable to corruption, such as public procurement and tax administration. In particular, the evidence indicates that AI is generally implemented as a tool that enhances, rather than replaces, human capabilities. Despite the notable financial outcomes reported in the literature, the analysis also highlights recurring challenges, including poor data quality, a shortage of skilled personnel, and a significant gap between technological advancement and the development of legal and ethical frameworks. In conclusion, the effectiveness of AI in combating corruption is contingent on contextual factors like institutional maturity and data governance. Additionally, a comprehensive strengthening of government capacities is required to ensure that AI is employed in a transparent, ethical, and responsible manner.
Keywords: fight against corruption, artificial intelligence, public sector, government transparency.
La corrupción estructural que deteriora el desarrollo y la confianza en las instituciones puede ser enfrentada con la inteligencia artificial, una herramienta con capacidad de transformación para fortalecer la transparencia gubernamental en el sector público. El objetivo de la investigación es identificar y analizar las características de las prácticas más relevantes que utilizan inteligencia artificial para combatir la corrupción en el mundo. El enfoque metodológico fue cualitativo, descriptivo y comparativo, basado en un estudio de 19 iniciativas internacionales. Se halla una tendencia mundial a utilizar la inteligencia artificial a través del machine learning supervisado, principalmente en áreas de alta vulnerabilidad como las compras públicas y la administración tributaria. Se observa que hay una implementación de la inteligencia artificial como una herramienta que complementa las capacidades humanas, no como un reemplazo. Aunque se han presentado resultados financieros significativos, el análisis evidencia desafíos comunes como la deficiencia en la calidad de los datos, la falta de personal capacitado y una notable brecha entre el avance tecnológico y el desarrollo de marcos jurídicos y éticos. En conclusión, la eficacia de la inteligencia artificial depende de factores contextuales como la madurez institucional y la gobernanza de datos, y necesita una reforma integral de las capacidades estatales con el fin de garantizar un uso transparente, ético y responsable.
Palabras clave: combate a la corrupción, inteligencia artificial, sector público, transparencia gubernamental.
Corruption is one of the main obstacles to community development because it affects all spheres of society (
Corruption undermines government transparency, the efficiency of public spending, and citizens’ trust in institutions (
In this context, Artificial Intelligence (AI) has been a useful tool for transforming public administration (
Although initiatives implementing AI to promote transparency have increased, there is still no systematic comparative study of these international experiences that identify their characteristics, similarities, and differences. The absence of a systematic approach makes it more difficult to provide transferable lessons and limits the opportunity to replicate successful models adapted to different institutional contexts.
Technopolitics in the public sector—the strategic and systematic use of digital systems for the organization, communication, and transformation of collective political action (
On the basis of these considerations, this article addresses the following question: What are the defining characteristics of the most relevant practices that employ AI to fight against corruption in the public sector worldwide? To answer this question, the study examines existing initiatives. It recognizes that their success depends not only on technological development (
This study conducts a comparative review of international practices. It analyzes key aspects, including the type of AI employed; the name, main characteristics, and legal and ethical framework of the initiative; the degree of technological maturity achieved; and the type of institutional function that is replaced or automated. The article aims to provide a clear and organized overview of the use of AI in public administration, to expand knowledge about its implementation, and to provide comparative evidence on the main characterizing elements.
This section establishes the theoretical and conceptual framework that brings together three areas of study for government modernization: government transparency, the fight against corruption, and the use of AI in the public sector. According to various investigations, greater transparency is critical to reducing corruption because it strengthens accountability processes (
Government transparency is one of the fundamental pillars for strengthening democracy and governance, ensuring access to public information about government activities, decisions, and resources (
Following
Transparency should not be viewed as an end in itself but as part of a comprehensive approach to open government that fosters citizen participation and collaboration among different sectors of society (
As mentioned above, digital platforms and open-data portals have transformed the government’s management of public information (
Corruption can be understood as behavior motivated by the desire to obtain private gain at the expense of the public good, in violation of established norms (
Bribery, extortion, nepotism, and influence peddling are some of the forms of corruption in administrative and political spheres, affecting the integrity and efficiency of public institutions.
This equation implies that acts of corruption increase if public officials have exclusive power over key decisions (monopoly), act freely without clear rules (discretionary power), and are not subject to oversight mechanisms and legal consequences (accountability). Similarly, corruption can be understood as a hidden transaction between rational agents seeking to maximize their benefit.
Furthermore, this problem is associated with decision-making regarding public spending and the diversion of resources toward projects that are less necessary but easier to manipulate (
Corruption remains a serious and widespread problem worldwide. In this regard, the Corruption Perceptions Index evaluates 180 countries and considers factors such as the misappropriation of public funds, the use of public office for personal gain, the integrity and independence of the judiciary, and bribery (Transparency International U.S., 2025). According to the 2024 index, more than two-thirds of countries score below 50 out of 100, indicating high levels of perceived corruption in the public sector globally (
The fight against corruption has become a global priority, with institutions such as the OECD and the World Bank leading the effort.
However, existing government measures remain largely reactive, focusing on punishing acts of corruption rather than working to prevent them, despite the persistence of corruption in countries with severe penalties (
AI in the public sector has emerged as a tool with significant transformative potential, shaping the way governments design and implement public policies, administrative management, service delivery, and interaction with citizens (
In various areas of public administration—such as tax collection, contract oversight, the justice system, public safety, and the management of social programs— AI is playing a key role in improving the responsiveness and accountability of public administrations (
Furthermore, this technological advancement enables the real-time analysis of large amounts of data, the identification of risk patterns, the automation of complex processes, and the provision of predictive information that strengthens evidence-based decision-making within the public sector (
The use of AI in the public sector is not without its challenges. One of the main risks is the lack of transparency in algorithms, known as the “black box” dilemma, in which automated systems make decisions without public officials or citizens clearly understanding how (
Another challenge is the protection of personal data. Since many AI models require access to large amounts of sensitive information, including financial, biometric, or health data, regulatory frameworks are needed to ensure privacy, non-discrimination, and information security (
The institutional capacity gap poses a key challenge to the effective adoption of technologies such as AI (
AI offers opportunities and challenges in the fight against corruption across various sectors, as it can process large amounts of data to detect fraudulent behavior, flag suspicious transactions, and identify irregularities in public procurement (
Therefore, the implementation of AI entails risks, such as biased algorithms, the potential manipulation of data for personal gain, and the possibility of concealing corrupt practices rather than exposing them (
From a technopolitical perspective, AI should be implemented strategically to foster organization and collective action in the oversight of public power (
AI in technopolitics facilitates a transition from a system of oversight with fixed timeframes to a process of constant and permanent monitoring (
Based on technopolitics, algorithms and data do not belong solely to large organizations and/or companies but are managed as a public good (
This study adopted a qualitative, descriptive, and comparative approach, which is suitable for examining complex and multidimensional phenomena such as the use of AI in government contexts. This method is also appropriate for describing and comparing existing experiences to facilitate an understanding of the strategies, technologies, and outcomes of various international initiatives
A literature review was conducted based on the collection, systematization, and analysis of relevant information obtained from reliable and specialized sources (
A systematic search (
Nineteen countries with varying levels of economic and technological development that have implemented notable initiatives using AI to detect, prevent, and combat corruption in the public sector were selected. The countries are Canada, the United States, Mexico, Colombia, Denmark, Spain, France, Italy, the United Kingdom, China, South Korea, India, Singapore, Ghana, Kenya, Nigeria, South Africa, Australia, and New Zealand. This diversity of countries aims to provide a global perspective on the different realities and institutional contexts, in order to identify significant similarities and differences.
Table 1 organizes and systematizes the characteristics of the practices identified in the reviewed texts, describing the variables that allow for the recognition of the most relevant elements of each initiative. This table was used as a tool for coding and analyzing the information, with data collected carefully to ensure quality and accuracy (
Variable | Description |
| Country | Location where the initiative is developed and implemented |
| Name of the initiative | Official or common name of the AI -based program, platform, or project |
| Year | Date of implementation or launch |
| Type of AI used | Technology classification, e.g., machine learning, natural language processing, predictive algorithms, big data |
| Sector of application | Areas of application: public procurement, tax administration, audits, citizen participation, monitoring and evaluation |
| Features | Overview, scope, and functionalities of the initiative and the technology used |
| Results | Reported impacts in terms of improved transparency, administrative efficiency, detection or reduction of acts of corruption, or increased citizen participation |
| Key challenges | Major obstacles identified, such as technological barriers, institutional resistance, and legal, ethical, or social constraints |
| Maturity level | Current stage of the initiative’s development, ranging from pilot projects to full-scale, widespread implementation |
| Legal or ethical framework | References to standards, regulations, or policies that support, govern, or restrict the use of AI in the specifict context |
| Sources | Documents, articles, reports, or websites from which the information was obtained |
| Who is it replacing? | Functions, processes, or people that are being replaced or assisted by technology |
This study was based on a process of systematization and analysis of data obtained from the literature and official documents for each of the selected cases (
This methodology facilitates the systematization of knowledge by bringing together topics related to the use of AI in the public sector, government transparency, and the fight against corruption.
This section presents a descriptive analysis of various international initiatives that employ AI to combat corruption. The study conducts a comparative analysis of leading countries on each continent, based on the similarities and differences identified with respect to the following characteristics: the type of AI used, the sector in which it is applied, what the AI is intended to replace, the legal or ethical framework governing it, the maturity level of the initiative, as well as the results obtained and the challenges involved in adopting this technological development (see Tables 2 through 6).
Variable | Colombia | United States | Canada | Mexico |
| Name of the initiative | VIGIA | Analytics Program | Fintrac | Plataforma Digital Nacional (National Digital Platform) |
| Year | 2021 | 2024 | 2023 | 2022 |
| Type of AI used | Supervised machine learning | Supervised machine learning | Supervised machine learning | Supervised machine learning |
| Sector of application | Public procurement | Federal payments | Financial intelligence | Public procurement |
| Features | Early warning system for public procurement | Analysis of federal transactions and detection of fraud in real time | Analysis of millions of bank transactions in real time and detection of patterns of money laundering and illicit financing | Identification of patterns of corruption risk in public contracts, focusing on the buyer-supplier relationship |
| Results | Improvement of contract oversight and early detection of irregularities | Prevention and recovery of over USD 4 billion in fraud | Increase in reports of suspicious transactions (from 114,422 in 2015 to 560,858 in 2022–2023) | Strong performance in identifying irregular contracts; a model applicable internationally |
| Key challenges | Lack of trained personnel and limited data quality | Data quality and availability; interagency coordination | Handling of large volumes of data and protection of rights | Quality and consistency of public data |
| Maturity level | Active | Operational and expanding | Operational and expanding | Pilot |
| Legal or ethical framework | Protection of personal data in accordance with Colombian law | Implementation based on Executive Order 14110 and standards related to AI, transparency, governance, and data protection | Money Laundering Act and Fintrac’s regulatory authority aligned with framework for responsible AI | In the process of alignment with ethical principles of transparency and data protection |
| Sources | ||||
| Who is it replacing? | It replaces human analysts who manually review contracts, databases, or potential cases of corruption involving public procurement. | It partially replaces traditional human auditors who conducted case-by-case reviews. | It replaces human financial analysts who previously analyzed suspicious transaction reports manually. | It complements the work of traditional oversight and regulatory institutions. |
Variable | Italy | France | Denmark | United Kingdom | Spain |
| Name of the initiative | Autorità Nazionale Anticorruzione (ANAC) | AntiCorAI | Social AI Monitoring System | Public Sector Fraud Authority (PSFA) | Estatal de Administración Tributaria (State Tax Administration) |
| Year | 2023 | 2023–2025 | 2015 | 2023 | 2024 |
| Type of AI used | Embedded AI and blockchain | Supervised machine learning | Supervised machine learning and data mining | Supervised machine learning | Supervised machine learning |
| Sector of application | Public procurement | Public finance | Social welfare | Fraud in public payments | Tax audit |
| Features | Automation of the public procurement process using AI and blockchain to prevent corruption | Analysis of tax data and transactions, automatically detecting tax fraud and anomalies | Use of up to 60 AI models to analyze administrative data, detecting fraud in welfare payments before they are processed | Use of the Snap platform to detect fraud networks in public payments through network analysis and machine learning | Analysis of tax data and transactions, improving the identification of taxpayers at risk of fraud |
| Results | Adoption of a new AI-based procurement code | 50% of fraud investigations initiated by AI | Recovery of EUR 23 million with an annual budget of 3.1 million | Identification of over EUR 140 million in improper payments | Improved selection of high-risk cases and increased debt recovery |
| Key challenges | Lack of empirical evidence on actual impacts | Protection of sensitive data | Risk of mass surveillance, algorithmic discrimination, protection of sensitive data, and ethnic bias | Data quality, interagency coordination | Risk of algorithmic opacity |
| Maturity level | Initial phase | Operational and expanding | Operational and expanding | Operational and scaled | Operational and scaling |
| Legal or ethical framework | Legislative Decree 36/2023; algorithmic oversight pursuant to Law 190/2012 | National data protection regulations and ethical principles for AI in the French treasury | National and European privacy regulations | Framework of the Government Counter Fraud Functional Strategy | Based on the National AI Strategy (2020), Royal Decree 817/2023, and future oversight by AESIA |
| Sources | |||||
| Who is it replacing? | It does not replace regulatory authorities; it automates and optimizes the analysis and monitoring of bidding processes and contracts. | It partially replaces manual reviews and traditional analyses of large volumes of tax data. | It partially replaces repetitive tasks such as the manual review of cases and verification of eligibility. | It partially replaces manual audits and traditional analyses in the detection of fraud in payments and procurement. | It partially replaces traditional manual processes for reviewing and analyzing tax returns. |
Variable | South Korea | India | Singapore | China |
| Name of the initiative | Koneps | Government e-Marketplace (GeM) | GovTech | Communist Party of China (CPC) Surveillance System |
| Year | 2006 | 2016 | 2023 | 2012 |
| Type of AI used | Supervised machine learning | Machine learning and deep learning | Machine learning and data mining | Supervised machine learning |
| Sector of application | Public procurement | Public purchasing | Public purchasing | Government oversight |
| Features | Use of machine learning to analyze contracts and detect collusion in public tenders through predictive indicators | Analysis of patterns in public purchasing and detect potential fraud or collusion | Correlation of data on human resources, supplier payments, and family relationships to detect the risk of fraud in public purchasing | Analysis of multiple databases (banks, real estate, transactions) and detect early signs of corruption among local officials |
| Results | Estimated savings of USD 6.5 million per year | Efficient purchases amounting to USD 65 billion between 2024 and 2025 | Regular identification of suspected cases | Identification of 8,721 officials involved (2012–2019) |
| Key challenges | Risk of fraud in less visible stages | Industrial data quality | Integration of sensitive data | Lack of algorithmic transparency |
| Maturity level | Operational | Operational | Pilot expansion | Operational |
| Legal or ethical framework | Legal requirements and transparency frameworks in accordance with the OECD | Digital government policy; currently incorporating blockchain and explainable AI; explainable AI government guidelines | Part of the Smart Nation initiative and GovTech policies, with internal data protection rules and human oversight of alerts | Backed by the party’s central disciplinary committee; internal oversight operation with little external transparency |
| Sources | ||||
| Who is it replacing? | It partially replaces manual and decentralized public procurement management processes. | It partially replaces manual processes for supplier evaluation, contract review, and risk management. | It partially replaces manual tasks such as contract review, supplier evaluation, and risk analysis. | It complements the oversight and monitoring efforts of the disciplinary bodies of the CPC. |
Variable | Kenya | South Africa | Nigeria | Ghana |
| Name of the initiative | KEBS-AI | Open Contracting Data Standard | Nigerian Integrity Portal | Open Data Initiative |
| Year | 2024 | 2023 | 2024 | 2023 |
| Type of AI used | Machine learning, data mining, and natural language processing | Machine learning and network analysis | Machine learning, data mining and natural language processing | Machine learning, data mining and natural language processing |
| Sector of application | Tax administration | Public bidding | Public purchasing | Public oversight |
| Features | Analysis of tax data, transactions, and imports to detect tax evasion and fraud | Analysis of public purchasing using machine learning and network analysis, flagging potential fraud and collusion in bidding processes | Use of AI, data mining, and analysis of citizen complaints to detect fraud in public procurement | Use of AI, data mining, and automated citizen reporting to detect fraud in public procurement and improper payments |
| Results | Recovery of USD 30 million between 2023 and 2024 | Detection of 30,000 fraudulent tenders over three years | Recovery of USD 4 million in less than 24 hours | Reduction of corrupt practices in public procurement |
| Key challenges | Internal sabotage and data privacy | Cross-platform interoperability | Fragmentation of public data | Institutional resistance |
| Maturity level | Pilot expansion | Pilot expansion | Pilot scaling | Pilot expansion |
| Legal or ethical framework | Kenya Revenue Authority (KRA) Act, Chapter 469 | Public Procurement Act 28/2024, Public Finance Management Act (PEMA) / Municipal Finance Management Act MEMA | Independent corrupt practices and other related of offences commission | Public Procurement Authority Act and Whistle-Blower Act (2006) |
| Sources | ||||
| Who is it replacing? | It replaces manual methods of auditing, oversight, and detection of tax evaders. | It complements the work of oversight and public purchasing. | It complements traditional civil registries and siloed government databases. | It complements government audit and control agencies. |
Variable | Australia | New Zealand |
| Name of the initiative | Robodebt Scheme | Algorithm charter for Aotearoa New Zealand |
| Year | 2025 | 2025 |
| Type of AI used | Supervised machine learning | Supervised machine learning |
| Sector of application | Social security and tax administration | Public sector |
| Features | Detection of tax fraud and improper payments in real time by analyzing large volumes of transaction data | Establishment of a framework for responsible AI to ensure its ethical, transparent, and citizen-centered use |
| Results | Prevention of USD 2.5 million in goods and services tax fraud | Adoption in 37 of 50 government agencies |
| Key challenges | AI governance in development | Lack of binding policies |
| Maturity level | Advanced pilot | Training/development |
| Legal or ethical framework | Australian National Audit Office (ANAO) guidelines; Australian Public Service Commission (APSC) on the responsible use of AI | In line with the OECD’s model; Privacy Act, Bill of Rights Act, and Human Rights Act |
| Sources | ||
| Who is it replacing? | It partially replaces the manual review of improper payments in social security programs. | It complements public sector agencies in making data-driven decisions. |
The mapping of these initiatives reveals a growing trend toward the adoption of AI as a tool for combating corruption. A comparative analysis across continents shows that, while regional differences exist, the application of this technology worldwide follows certain patterns. Efforts are concentrated in highly vulnerable sectors of public administration, such as public procurement and purchasing, tax administration, and the management of federal or social welfare payments (
Supervised machine learning is the most widely used AI technology (
A key finding is that AI is not used as a substitute for humans but rather as a tool to complement the capabilities of government institutions (
The level of maturity of these initiatives varies significantly across regions (
The reported quantitative results are one of the most important indicators of the use of AI in the fight against corruption. The United States’ Analytics Program has prevented and recovered over USD 4 billion in fraud (
The implementation of AI faces common challenges worldwide. The main obstacle identified in nearly all initiatives is the quality, availability, and consistency of data. The effectiveness of machine learning models depends directly on the quality of the data used to train them (
With regard to legal and ethical frameworks, most initiatives operate under existing regulatory systems and laws, such as those governing personal data protection. However, the development of regulatory frameworks specifically designed for the adoption of AI in the public sector remains a challenge in many countries (
Findings on technopolitics reveal that the use of AI is a common and evolving strategy for combating corruption, primarily in areas of high institutional vulnerability, such as the management of social programs, tax administration, and public procurement. Particular characteristics and adjustments of AI strategies depend on each country’s context and circumstances. Quantitative data reveal significant economic impacts in several countries. However, the technopolitical potential of AI still faces persistent challenges related to the quantity and quality of data, a shortage of specialized experts, and the lack of specific legal and ethical frameworks.
AI must avoid the arbitrariness and opacity of closed algorithms so that technopolitics can work against corruption (
This study identified the most significant initiatives worldwide that utilize AI to fight against corruption and promote government transparency. The analysis was conducted through a literature review employing a descriptive, qualitative, and comparative approach to selected cases, compiling relevant information from 19 country-level initiatives.
The results reveal that technological developments in AI are useful for combating corruption, although its implementation alone does not solve such a complex problem. According to the comparative analysis, the success and sustainability of these initiatives are linked to factors such as institutional maturity, the quality of data infrastructure, and the presence of governance frameworks. In addition, the findings show that AI does not accelerate processes, that its functionality depends on the existing conditions under which it is implemented, and that it helps enhance people’s capabilities and skills rather than completely replacing them (
AI is used in most projects to automate anomaly detection across various datasets, enabling strategic analysis and complex decision-making. In this context, the challenge for public organizations is to train their workforce to understand and act on the results generated by algorithms. Furthermore, given that data are the most important element for training AI models, some limitations to its potential must be addressed. These include the lack of interoperability between agencies, the diversity of public organizations’ databases, and the poor quality and standardization of information. In this regard, rather than investing in sophisticated algorithms on a large scale, strategic efforts are needed in data governance and the creation of a comprehensive digital infrastructure.
The study reveals that global digital governance is uneven due to differences in the maturity and scale of AI applications across countries. While Europe and North America have expanding systems with quantifiable financial results, initiatives in Africa and Latin America are in pilot phases and face structural barriers. Additionally, there is a significant gap between the rapid technological advancement of AI and the creation of legal and ethical frameworks to regulate it. Most initiatives are addressing the lack of transparency and algorithmic biases with outdated regulatory frameworks ill-suited for this purpose. This absence of specific regulations for using AI in the public sector can generate uncertainty and mistrust among citizens toward the government. In conclusion, disparities in institutional maturity and differences in regulation and infrastructure show that AI’s potential to combat corruption is shaped by political, economic, and ethical contexts.
This study has limitations that must be acknowledged. The analysis is based on a descriptive mapping of initiatives drawn from public sources, which limits an in-depth causal assessment of the impact of AI on reducing corruption levels. Future research should focus on measuring the impact of AI, as well as on qualitative analyses of adaptation processes in public organizations and the institutional resistance that arises during the implementation of these technologies. The comparative analysis presented here sought to offer an overview of global initiatives in this area, instead of generalizing solutions. This article highlights significant progress in the implementation of AI to fight against corruption. However, greater standardization and impact assessment are still needed because many of the projects are in pilot phases and lack independent evaluation.
The findings suggest that technopolitics in the analysis of AI for combating corruption and promoting government transparency is a process of institutional change. When strategically integrated into the dynamics of public action, this shift can enhance the government’s capacities in terms of detection, monitoring, and accountability. AI provides unprecedented tools for surveillance and data analysis, but it requires a transformation of government capabilities, processes, and structures in the areas of data governance, human capital development, and regulatory frameworks. Thus, the challenge lies not only in programming the algorithm but also in redefining the public sector
The authors declare no conflicts of interest.
All authors contributed equally to this article and share joint responsibility for its content.
The authors disclose that they used the Gemini generative AI tool exclusively to correct spelling, punctuation, and syntax during the manuscript preparation phase. They also guarantee that the tools used were employed in an ethical and honest manner.