Malffunctions under the lens
In this context, the use of artificial intelligence should also be included in risk analysis and to help officials identify the possible situations in which the greatest danger of conduct in breach of customs regulations lurks. Thus, the Agency’s ‘in-depth work on the role of artificial intelligence’ is moving forward,’ explained Director Alesse. And he added: “In recent years, the centrality of this technology in the customs sphere – with particular reference to e-commerce, a sector characterised by a very high volume of shipments – where it offers the possibility of detecting fraudulent patterns and anomalies, thanks to techniques based on machine learning, has become increasingly evident.
Scanner images
The system used by Customs has been enabled to automatically recognise, for example, anomalies in scanner images of cargo. In practice, as a consignment passes under the scanner, the algorithm compares the image with thousands of patterns of counterfeit or concealed goods, instantly alerting officials to potential anomalies. Ai, based on explainability verification, offers the possibility of detecting fraudulent patterns and anomalies that would escape manual checks or those based solely on static rules, thanks to various machine-learning-based techniques that improve the accuracy of non-intrusive checks, especially in e-commerce, where the volume of shipments is very high.
Focus on small packages
E-commerce is inevitably a special watchdog because various illegal situations pass through online purchases and sales. Also for this reason, the 2026 strategy looks with great attention at e-commerce and in particular at low-value shipments, i.e. those below 150 euro, arriving from non-EU countries. Apart from the issue of the EUR 2 contribution introduced by the manoeuvre and the tariffs approved by the EU Council coming in on 3 July, the exponential growth of this sector has led to an increase in fraud attempts linked to under-invoicing and the evasion of tariffs. This is shown by the numbers in recent years:
- in 2023 out of more than 29 thousand customs declarations analysed (track H7) a 8.3 per cent discrepancy rate was found;
- in 2024 the volume of analyses rose to about 45 thousand declarations, with an increase in the discrepancy rate to 10.5 per cent.
Big data professionals
Technology alone is not enough. That is why the Agency has strengthened its staff by hiring highly qualified professionals and experts in big data management, the application of artificial intelligence and the development of advanced technologies.
A full-range support
