Mesterséges intelligenciával támogatott innovatív korrupcióazonosító rendszer alapvetései a közigazgatásban

Effectively combating corruption is one of the most pressing challenges of our time, particularly in the realm of public administration. The Hungarian government has been committed for decades to reducing all forms of corruption, protecting the European Union's financial interests, and ensuring the lawful use of EU budgetary resources. The current guiding document for these efforts is the National Anti-Corruption Strategy (hereinafter NKS) for the period from July 1, 2023, to December 31, 2025. This strategy was developed within the framework of the conditionality procedure and the Recovery and Resilience Plan. As part of this initiative, a risk-based evaluation of job positions and roles has been conducted across the entire public administration workforce. Similar to earlier measures, this evaluation serves as the foundation for analyses that determine the corruption exposure of various positions and roles. Aim: is to achieve tangible results that go beyond research studies and recommendations. The goal is to develop an innovative tool that public administration leaders can use to effectively identify corruption risks and uncover corruption-related activities, ideally at an early stage, before criminal acts are completed, thereby avoiding criminalization. This tool not only supports the functioning of individual organizations but also leverages artificial intelligence and advanced statistical and data analysis tools. It provides leaders and auditors with a proactive and targeted means to combat corruption, enhancing institutional transparency and public trust. Methodology: The most effective way to carry out this mission is to design and develop technology resulting in an Innovative Corruption Detection System. This system employs algorithms designed to detect corruption patterns and anomalies in procedures. It also uses artificial intelligence to analyse the content of processes and related documents, identifying, verifying, and filtering potentially suspicious activities by corruption type. Furthermore, it automatically flags events recommended for investigation. The system operates as a database-driven network software with process-controlled user architecture. Its main functionalities include identifying anomalies in discretionary decisions, detecting various types of economic corruption, initiating and closing investigations, and producing results reports. These reports feature predictive models indicating trends and provide a status registry ('criminal record') for maintaining oversight. Findings: The system allows the investigation of identified events even during ongoing procedures, enabling timely corrections and reducing the number of completed corruption cases. Key input points include risk assessments based on roles and positions, data from organizations, applicants, suppliers, beneficiaries, decision-makers involved in events, their registry affiliations, and event-related data and documents. By incorporating artificial intelligence, the system identifies statistical trends, functioning as a forecasting tool. This supports decision-making for preventive measures, significantly aiding organizational leaders and auditors in their work. Value: Beyond economic activities and events, this procedure can extend to other forms of corruption risk, such as nepotism or discretionary decision-making. Nationwide implementation would enable comparisons across sectors and institutions, identifying necessary intervention points. This approach not only focuses on regulatory and legal frameworks but also places greater emphasis on examining the human factor. Analyses could also evaluate how corruption trends evolve following the introduction of these innovations and their measurable impact on state expenditure. This will provide the Government with another effective tool to demonstrate its commitment to significantly reducing corruption, which can bring recognition at international level.