AML research shows transaction vs actor-level scoring impacts investigation queues

This paper investigates the impact of granularity in graph-based anti-money laundering (AML) systems for blockchain networks. Researchers evaluated whether scoring suspicious activity at the transaction level or the actor address level affects the composition of investigation queues. Using the Elliptic++ Bitcoin dataset, they found that transaction-level projections led to significantly different investigation queues compared to address-level scoring, impacting the efficiency and focus of AML efforts. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This research highlights how data granularity in AI-driven AML systems can significantly alter investigation outcomes, suggesting a need for careful design choices in financial compliance tools.

RANK_REASON Academic paper evaluating a methodology for graph-based AML systems.

Read on arXiv cs.LG →

• Elliptic++
• Bitcoin
• arXiv

• paper
• other