Chinese Police AI Spots Illicit Bitcoin with 89.4% Accuracy

Chinese authorities have already been active in crypto-related enforcement. In March 2025, prosecutors indicted 3,259 individuals for money laundering tied to virtual currencies and underground banking. On July 25, 2026, a Chinese court handled a case involving nearly 3 billion yuan, about $444 million, linked to online gambling debts. The new AI tool could accelerate such investigations.

The research concentrates exclusively on Bitcoin. Dr. Sun's team did not test other cryptocurrencies, reinforcing Bitcoin's historical role as the primary rail for illicit transactions. If deployed, the system could push criminals toward privacy-focused coins or off-chain methods, though the study does not address that directly.

For global law enforcement, the model offers a template that could be adapted to other blockchains. The open publication of the methodology means agencies outside China could replicate the approach. Whether they will depends on data access and regulatory frameworks, researchers said.

Existing detection tools typically rely on static graph analysis or rule-based heuristics. The new framework's use of a memory module and LLM allows it to adapt to evolving laundering techniques, the researchers said. The high precision reduces false positives, which is critical for resource-constrained agencies. The low recall, however, means many illicit transactions still go unnoticed, leaving room for further refinement.

A tool that catches two-thirds of illicit Bitcoin transactions is a start. The remaining blind spots leave room for further refinement.