China develops AI system to detect illegal crypto transactions

Researchers at the Chinese Public Security University have created an artificial‑intelligence system that can identify illicit cryptocurrency activity with roughly 89‑90% overall accuracy. The platform combines a dynamic graph neural network, a memory module and a large language model to analyse transaction patterns, trace money flows and compare new cases with previously flagged illicit behavior.

The development is part of a broader crackdown on crypto‑related financial crimes. Chinese prosecutors announced in 2025 that indictments had been prepared against 3,259 individuals suspected of money‑laundering through virtual currencies, and the new AI tool is intended to give regulators a more sensitive, generalisable and explainable means of spotting such activity.