IOTA Forum on Tax and Customs Data Science “Data and AI – from Governance to Guided Analytics”

With the increasing volume of transaction data and the complexities of international trade, tax authorities in both EU and non-EU countries can significantly benefit from utilising tax and customs data as a key resource for their analytical endeavours. This approach can enhance compliance, optimise risk management, and provide tailored services to taxpayers and traders alike. Looking ahead, this initiative will help boost the effectiveness and efficiency of data governance and management within tax authorities and also align with the broader goal of improving global tax and customs compliance through innovative, data-driven methodologies.

The meeting kicked off with plenary sessions that outlined the forum’s background, goals, and objectives, along with an open discussion on the draft mandate. Participants also delved into the forum's operational aspects, including agenda formulation, steering group composition, and planned sub-groups and projects. These introductory sessions were facilitated by IOTA Experts Jonathan Heath and Massimo Morarelli.

The first day of the event then moved on to focus on the approaches and experiences of two IOTA members, Belgium and Norway, in relation to Data Governance and Management, particularly in the context of enforcing data governance.

• Belgium: “Enforcing Data Governance: From Framework to Operational Excellence” co-presented by Wim Bonneux & Maarten Vanderjeught;
• Norway: “The Norwegian experience” presented by Knut Magne Watne.

The Virtual Group Discussions also allowed participants to discuss their own experiences of data governance and enforcement. Participants had the opportunity to share their administration’s approach and explore each other’s methods.

On the second day, the focus was on using AI in analytics, featuring four presentations from IOTA members demonstrating different ways AI has been used for risk assessment and fraud identification. The following country presentations reflected on this topic. The countries that shared their experiences were:

• Italy: “Categorising e-Invoice Data through Local LLMs” presented by Francesco Ascione;
• France: “Foncier Innovant: DGFiP’s AI-based Approach to Detect Real Estate Tax Fraud” presented by  Victoire Daher;
• Sweden: “Automated Risk Detection at Scale: How Sweden’s Tax Agency Uses AI to Target High-Risk Deductions” presented by Marcus Bräutigam;
• Austria: “Autoencoders: Anomaly Scores as Features in Supervised Models for Different Tax Areas” presented by Konstantin Posch.

Like the previous day, attendees engaged in group discussions about the use of AI in risk analysis, sharing some experiences of how AI has been used within their own administrations and considering some of the risks and benefits.

The new Forum concluded with an interactive Q&A session where participants could raise any additional questions about what they had heard during the two days of the event and share their thoughts on what actions should arise from the meeting.

Although the Forum primarily focused on tax data, many IOTA members also operate as combined tax and customs authorities, where customs data plays a crucial role in combating tax fraud. Therefore, the broader long-term aim is to explore both tax and customs data. All materials and presentations are now available online on the event’s webpage, and the video recording will be made accessible soon. This new IOTA Forum was launched and facilitated by IOTA International Taxation Experts Jonathan Heath and Massimo Morarelli.