AN GARDA SIOCHANA

The FINTERCEPT project will develop and operationalise an explainable, and sustainable artificial intelligence (AI) framework for counter-terrorist-financing (CTF) intelligence, transforming fragmented analytical capacities into a unified, interoperable European ecosystem. The project introduces a distributed analytical twin—a digital representation of Europe’s financial-intelligence landscape—that learns, adapts, and operates across multiple jurisdictions without centralising sensitive data. This framework integrates advanced computational modelling, lawful data harmonisation, and multi-layered explainability to identify hidden typologies and transactional behaviours indicative of terrorist financing. FINTERCEPT’s AI architecture combines graph neural networks for networked entity analysis, anomaly-detection models for behavioural irregularities, and large language models capable of processing unstructured financial narratives and open-source intelligence. By linking quantitative and semantic reasoning, the system enables cross-domain interpretation and human-understandable insights, ensuring reliability, transparency, and auditability under the GDPR, AML directives, and the forthcoming EU AI Act. Privacy-preserving machine learning techniques, such as encryption and secure multi-party computation, safeguard data sovereignty while supporting collaborative intelligence across jurisdictions. The infrastructure will be validated through multi-level demonstrations in national and cross-border operational environments involving financial institutions, FIUs, and law-enforcement authorities. These pilots will evaluate analytical accuracy, interoperability, and legal admissibility, generating evidence for large-scale deployment. FINTERCEPT’s long-term ambition is to establish a lawful, trustworthy, and human-centric AI paradigm for European financial-intelligence analysis, strengthening institutional resilience, enhancing cross-border cooper

EC contribution:

EUR 139k

Start:

2026-11-01

End:

2029-10-31

Status:

SIGNED

Scheme:

HORIZON-RIA

Call:

HORIZON-CL3-2025-01

View on CORDIS (DOI 10.3030/101309470)