The use of artificial intelligence is becoming an important area of technological development for anti-corruption agencies. AI tools are being deployed both to automate employees’ routine tasks and to support the prevention, detection and investigation of corruption offences. In particular, these technologies enable agencies to process large volumes of information, analyse complaints and reports, identify corruption risks and support investigations.

Poland

In October, Poland’s Central Anti-Corruption Bureau (CBA) announced that it had received the Best Initiative Award 2026 at the 25th Annual Conference of the European Partners against Corruption (EPAC) and the European Contact-Point Network against Corruption (EACN) for developing the M-AI-A artificial intelligence platform.

The platform integrates various AI tools to support CBA personnel in their daily work. Its capabilities include text analysis and document summarisation, automated translation, speech recognition, image processing and a virtual assistant. Bringing these tools together in a single platform is intended to streamline information processing and reduce the time spent on routine tasks. However, information on the extent of the platform’s deployment and its results has not yet been published.

M-AI-A is not the CBA’s only AI-based development. The agency previously introduced ETNA, a system designed to process reports from citizens concerning suspected corruption offences.

As reported during an EPAC/EACN webinar in April 2025, the system uses large language models (LLMs) and other AI tools to process incoming information. In particular, these technologies make it possible to digitise materials that are difficult to process using conventional software and to automatically analyse the content of reports.

CBA specialists are also working to expand ETNA’s capabilities. Planned features include classifying reports according to risk level and urgency, as well as identifying links between incoming reports, known corruption schemes and individuals under investigation.

South Korea

In the Republic of Korea, the Anti-Corruption and Civil Rights Commission (ACRC) is introducing generative artificial intelligence into its handling of citizens’ complaints and petitions.

On 5 February 2026, the commission launched new AI-powered services on the e-People platform, which enables citizens to submit complaints and petitions to government agencies.

The system includes three main tools:

1. The first assists in drafting responses to complaints and petitions. AI analyses applicable laws and regulations, previously processed cases and internal guidance materials before generating a draft response for the responsible official.
2. The second automatically identifies and groups submissions with identical or similar content, allowing them to be processed together and reducing repetitive work.
3. The third analyses the meaning and context of submissions. Unlike conventional keyword-based searches, it can identify related issues and use the resulting insights to support proposals for improving public administration.

During the initial phase, the new services were introduced on a pilot basis in four central and local government bodies, including the Ministry of Food and Drug Safety, the Ministry of Land, Infrastructure and Transport, and the municipal administrations of Incheon and Siheung.

Unlike Poland’s ETNA, which is designed to process reports of suspected corruption offences, the Korean system covers a broader range of administrative complaints and petitions. Its implementation demonstrates how AI can help automate interactions between government agencies and citizens and facilitate the analysis of information submitted by the public.

United Kingdom

In the United Kingdom, the Serious Fraud Office (SFO) uses machine learning technologies to analyse electronic materials in criminal investigations.

The agency uses OpenText Axcelerate, a software platform capable of processing large volumes of digital information. In particular, Technology Assisted Review (TAR) helps identify documents that may be relevant to criminal proceedings. After being trained on documents previously reviewed by specialists, the system analyses the remaining materials and ranks them according to their likely relevance to an investigation. SFO personnel make the final decisions on the relevance of the documents.

A pilot project demonstrated that TAR could make the review of materials subject to disclosure to the defence up to 40% faster than traditional methods. Following the pilot, the SFO announced plans to expand the use of the technology.

The introduction of digital tools has also been accompanied by efforts to verify their reliability. In February 2025, the SFO reported that it had resolved a technical issue in Axcelerate that affected the display of certain words in processed documents. The agency also conducted a review of its previously used Autonomy Introspect system due to issues with its search functionality that may have resulted in relevant materials being missed in criminal cases.

Hong Kong

In Hong Kong, the Independent Commission Against Corruption (ICAC) is developing its own artificial intelligence tools to support corruption investigations.

In August 2026, the commission unveiled Tianma, an AI-powered system designed to integrate and analyse information from multiple sources. The system is also intended to improve the efficiency of digital forensic examinations and financial investigations.

Development of Tianma began in 2025. In late August 2026, ICAC representatives demonstrated the system at an event organised by the Global Operational Network of Anti-Corruption Law Enforcement Authorities (GlobE Network) of the United Nations Office on Drugs and Crime (UNODC) in Vienna. The event brought together more than 130 specialists from approximately 70 anti-corruption, law enforcement and regulatory agencies around the world.

According to the ICAC, Tianma is intended to improve the quality and efficiency of investigations. The commission continues to enhance the platform, including through the integration of new technologies. However, quantitative results of its use have not yet been published.

Brazil

In Brazil, the Office of the Comptroller General (Controladoria-Geral da União – CGU) uses artificial intelligence primarily to analyse public spending, detect irregularities and assess integrity risks.

One of the agency’s best-known tools is Alice (Analisador de Licitações, Contratos e Editais), an automated system for analysing public procurement procedures and contracts.

Alice collects procurement data from government information systems on a daily basis, cross-references it with other databases and analyses procurement documents using text-processing technologies and artificial intelligence.

The system can identify indicators of potential irregularities, including inflated prices, restrictions on competition and collusion among bidders. When suspicious circumstances are detected, alerts are generated for government auditors, enabling them to conduct further checks and, where necessary, take action before contracts are awarded.

According to CGU estimates, the use of Alice has helped prevent procurement irregularities and generated public savings of more than BRL 11 billion since the system was created.

Alice is not limited to procurement procedures conducted by federal executive agencies. Its analytical results are also made available to state and municipal oversight bodies, provided that the relevant procurement procedures are published in information systems covered by Alice.

In addition to Alice, the CGU is developing other AI tools through its innovation laboratory, CGULab. These include Primus, designed to analyse integrity risks at the municipal level, as well as large language model-based solutions such as LIA, LLM2GOV and Eva no ePAD. These projects cover various areas of the agency’s work, ranging from analytical support for oversight activities to the automation of internal administrative processes.