Tunisia's customs authority is set to leverage artificial intelligence (AI) and machine learning to enhance risk management and monitoring efforts. The initiative, presented by the Ministry of Finance and the General Directorate of Customs, aims to analyze large datasets and identify high-risk shipments. This approach will enable the authority to focus its monitoring efforts on priority cases, while emphasizing that AI will serve as a tool to support decision-making, rather than replace human expertise.
A simulation model for machine learning in customs risk management was presented, based on 2025 data. The model demonstrated the potential to detect additional amounts of approximately 96.9 million dinars, with a detection rate of 49.23%. The model utilized 11 million observations and 27 initial variables, which were enriched with composite variables to reach 42 variables. Ultimately, 11 variables were selected to build the model, which employed a decision tree approach to facilitate interpretation and explanation of risk classification.
The customs authority's experience relies on integrating monitoring results into the model's retraining and periodic updates. This allows for a shift from static selectivity, based on predefined rules, to dynamic selectivity that benefits from data and previous monitoring results. In the tax sector, a project was presented to utilize big data and machine learning to improve tax monitoring targeting. The project aims to identify high-risk files and allocate resources accordingly.
The project also seeks to enhance the administration's ability to monitor the informal economy, unregistered individuals, and new forms of digital activity. Additionally, it aims to provide decision-making tools and combat complex fraud patterns. According to the Director General of the Information Center at the Ministry of Finance, the project is currently in the bidding phase, with an expected implementation period of one year.
The technical infrastructure will rely on local architecture and open-source tools, allowing for the experience to be applied to other areas of public finance. The General Director of Performance at the General Directorate of Performances highlighted that digitalization has reduced physical and repetitive tasks, freeing up time for monitoring and analysis. He noted that the number of remote declarants has reached 157,000, while 87,000 landlord declarations were submitted within the deadline.
The automated processing of landlord declarations has saved approximately 8,000 working days. The increasing volume of data generated by electronic invoicing and registered funds requires the development of administrative capacities for processing and analysis. AI can contribute to improving service quality, verifying data consistency, and linking transactions between various stakeholders.
A national symposium was organized to discuss the challenges and opportunities of AI in finance, taxation, and customs. The event, titled "Challenges of Artificial Intelligence and Fields of Intervention for the Ministry of Finance: The Role of Training Institutions," brought together experts to explore the potential of AI in enhancing public finance management.
Key points
- Tunisia's customs authority adopts AI and machine learning to target high-risk shipments and optimize monitoring efforts.
- The initiative aims to analyze large datasets and identify high-risk shipments, enabling the authority to focus its monitoring efforts on priority cases.
- The project is expected to enhance the administration's ability to monitor the informal economy, unregistered individuals, and new forms of digital activity.
