Synthetic AML/XAI Publishable Sample Dataset for a Compliance-Driven Explainable AI Architecture in AML/CFT Decision Support

This dataset contains fully synthetic sample data developed to support the architectural demonstration presented in the associated study on a compliance-driven explainable artificial intelligence (XAI) architecture for anti-money laundering and counter-terrorist financing (AML/CFT) decision support.

The dataset is provided exclusively as an illustrative research artifact and is intended to accompany the conceptual, traceability, explainability, and evidence-packaging components of the paper. It includes synthetic customer records, account master data, transaction-level events, illustrative rule definitions, rule-hit outputs, alert-level metadata, graph-ready transaction relationships, explanation objects, and a minimum evidence payload aligned with the proposed architecture.

The synthetic scenario operationalized in the dataset represents a simplified AML/CFT alert flow in which Customer A receives multiple near-threshold inbound transfers within a short time window, rapidly transfers funds to a newly established beneficiary, Customer B, and Customer B subsequently transfers funds to Customer C through a cross-border corridor. This scenario is designed to illustrate typologies such as structuring, layering, pass-through behavior, and monitored corridor transfers.

This dataset is intended solely for architectural illustration, transparency, and reproducibility support. It must not be used for benchmarking detection performance, evaluating predictive accuracy, or representing operational AML/CFT effectiveness in real financial institutions.

All records, identifiers, timestamps, entities, and values are artificially generated. No entry corresponds to any real individual, account, institution, or financial activity. The dataset contains no personal data, no confidential institutional data, and no proprietary operational information. It is therefore suitable for public dissemination as a synthetic research dataset.