CustomAI: AI technologies for enhancing capabilities, greater efficiency and effectiveness for customs

Description of the project

Customs authorities play a vital role in the global trade ecosystem – preventing the entry of illegal goods, safeguarding public safety and revenue, and ensuring the smooth flow of international commerce. However, as trade volumes continue to grow, balancing these responsibilities has become increasingly complex. Detecting illicit goods amidst millions of legitimate shipments is a mounting challenge, especially with limited human resources.

CustomAI brings together a multidisciplinary consortium of experts to develop an AI-driven toolkit designed to enhance the accuracy and efficiency of customs operations. The ultimate goal is drastically reduce false positives – cases where shipments are flagged for inspection despite being legitimate – while ensuring that high-risk cargoes are accurately identified.

By integrating non-intrusive, robust, and intelligent technologies, CustomAI will enable customs authorities to predict, detect, and select high-risk containers, parcels, and cultural goods for inspection with unprecedented precision.

At the heart of CustomAI lies the Virtual Customs Control Office (VCCO) concept – a digital environment encapsulating all the knowledge and processes needed to manage customs control for an artefact (e.g., container or parcel). The VCCO integrates a suite of cutting-edge technologies.

Industrial partners

Indra´s rol

INDRA participates in the CustomAI project as a beneficiary, together with its affiliated entities (ISS and IFT).

INDRA’s role focuses on integrating AI-based solutions into real customs operational environments, leveraging its Custos platform and expertise in security systems and decision support.

Key Contributions:

• Integration of AI-based risk analysis into customs systems
• Contribution to technical coordination and innovation management
• Development of user interfaces for customs officers
• Participation in pilot implementation and validation activities
• Contribution to AI-based decision support systems

Universities and technology centers

Tecnologies employed

1. AI-based risk anticipation: combining internal knowledge with external multilingual data such as manifests and declarations to identify only the most relevant cargoes for inspection.
2. AI-enhanced vapour-based detectors: applied selectively to containers flagged in the previous risk assessment stage.
3. AI-based x-ray analysis: for threat detection, streamlining the inspection process and involving human inspectors only when necessary.
4. Multimodal LLM Continual Learning: leveraging x-ray and camera imagery to continuously improve threat recognition models based on real customs data.
5. Blockchain-enabled data sharing: to ensure secure, transparent, and traceable supply chain communications.
6. Semantic NeRF-based representation: for creating and comparing 3D digital models of cultural artefacts, aiding in the detection of illegally exported cultural goods.

More information

This project, with file number 101226029, has been funded by European Union’s Horizon Europe Research and Innovation Programme, HORIZON-RIA Action, Topic HORIZON-CL3-2024-BM-01-05.