Exploring the In-Context Learning Capabilities of LLMs for Money Laundering Detection in Financial Graphs

TL;DR

This paper investigates using large language models to reason over financial graphs for money laundering detection, demonstrating their ability to identify suspicious activity and generate explanations in synthetic AML scenarios.

Contribution

It introduces a lightweight pipeline for LLM-based reasoning over financial graphs, showcasing potential for explainable AML analytics.

Findings

LLMs can emulate analyst reasoning in AML scenarios.

They can identify red flags and provide explanations.

The approach demonstrates potential for explainable financial crime detection.

Abstract

The complexity and interconnectivity of entities involved in money laundering demand investigative reasoning over graph-structured data. This paper explores the use of large language models (LLMs) as reasoning engines over localized subgraphs extracted from a financial knowledge graph. We propose a lightweight pipeline that retrieves k-hop neighborhoods around entities of interest, serializes them into structured text, and prompts an LLM via few-shot in-context learning to assess suspiciousness and generate justifications. Using synthetic anti-money laundering (AML) scenarios that reflect common laundering behaviors, we show that LLMs can emulate analyst-style logic, highlight red flags, and provide coherent explanations. While this study is exploratory, it illustrates the potential of LLM-based graph reasoning in AML and lays groundwork for explainable, language-driven financial crime analytics.