Agentic AI
New AI Agent Automates Detection of Cryptocurrency Money Laundering Across Blockchain Networks
Researchers introduced RISKTAGGER, an AI-powered system that combines language models with blockchain forensics to trace complex money laundering schemes in Web3 incidents with high accuracy.
Cryptocurrency investigators face a persistent challenge: when fraudulent funds move across blockchain networks, the money leaves a scattered trail of transactions that's difficult to piece together manually. Researchers have now developed an automated solution to this problem.
The RISKTAGGER System#
A new forensic analysis tool called RISKTAGGER merges large language model capabilities with blockchain investigation workflows. Rather than simply deploying a general-purpose AI system against raw transaction data, RISKTAGGER constrains the model's reasoning to evidence recovered directly from on-chain sources. The system acts as a guided agent: it reviews public incident reports for clues, then systematically reconstructs fund-flow pathways by analyzing cryptocurrency transactions while tagging accounts and activities by risk level.
The approach differs from existing Web3 anti-money laundering (AML) methods, which typically rely on manual analysis, heuristic rules, or graph-search algorithms and often produce only lists of flagged addresses without detailed reasoning or verifiable justification.
Real-World Performance#
Evaluating RISKTAGGER against five documented cryptocurrency incidents revealed strong performance across varying attack patterns and laundering architectures. In the primary test case (involving Bybit), the system identified 97.33% of addresses involved in the scheme and maintained 98.69% accuracy when expert analysts reviewed a sample of results. Results from the other four cases ranged from 95.24–100.00% recall and 91.27–100.00% precision.
What Makes Crypto Laundering Complex#
The investigation across multiple incidents uncovered why tracking illicit Web3 funds is so intricate. Money laundering in cryptocurrency ecosystems exploits several sophisticated tactics: short-cycle fragmentation (splitting funds across multiple transactions quickly), long-range laundering paths (dispersing funds over extended timeframes), interwoven DeFi services (mixing legitimate and illegitimate protocols), and deterministic denomination splitting (breaking transfers into predictable amounts to evade detection).
Why It Matters#
RISKTAGGER can recover the complete paths criminal funds take through blockchains, highlight the most dangerous accounts for investigators to prioritize, and compile its findings into human-readable forensic reports supported by public blockchain evidence. This bridges the gap between raw on-chain data and actionable intelligence for compliance teams and law enforcement.
