LLM Framework for Explainable AML Triage with Evidence Retrieval

A new arXiv preprint (2604.19755) proposes an explainable anti-money laundering (AML) triage framework that uses large language models (LLMs) with evidence constraints. The method combines retrieval-augmented evidence bundling from policy guidance, customer context, alert triggers, and transaction subgraphs; a structured LLM output contract requiring explicit citations and separating supporting, contradicting, and missing evidence; and counterfactual checks to validate decision coherence under minimal perturbations. The approach addresses risks of hallucinations and weak provenance in regulated workflows.