AI Model Spots Illicit Crypto Transactions with 89.4% Accuracy

Chinese researchers at the People's Public Security University have developed an AI system that can detect illicit Bitcoin transactions with 89.4% overall accuracy.

The team published their findings in May in the Journal of Intelligence and claimed that their framework outperforms existing mainstream baseline models for detecting dirty money flowing through blockchain networks.

The AI combines three technologies: dynamic graph neural networks, a memory module that stores historical patterns of illicit activity, and large language models. The researchers tested their system on the Elliptic Bitcoin dataset, which contains 203,769 transaction nodes and 234,355 edges connecting them.

The model achieved 89.1% precision and a 64.5% recall rate specifically for illicit transactions, meaning it flags something as suspicious about 89% of the time but only catches about 65% of all illicit transactions.