China develops AI to detect crypto laundering

TLDR

Chinas public security researchers have built an AI system that reportedly detects crypto-related money laundering with about 90 percent accuracy, tightening oversight of virtual asset flows in and out of the country.

1. The new framework uses AI and large language models to analyze blockchain data and flag laundering patterns like layering and structuring.
2. China is pairing strict crypto restrictions with sophisticated forensics, raising enforcement risk for mixers, OTC desks, and cross-border channels touching Chinese users.
3. Accuracy limits, false positives, and criminals own use of AI mean this will not end laundering, but it likely raises the bar and will influence global regulatory tech.

Deep Dive

1. How Chinas AI Laundering Detector Works

Researchers at Chinas National Police University have developed an AI-powered detection framework under the Ministry of Public Security to spot money laundering and economic crimes involving virtual assets.

The system combines pattern-recognition models with large language models (LLMs) to ingest blockchain transaction flows and related data, then identify anomalies consistent with laundering, such as rapid movement through multiple addresses (layering) or structured transfers designed to evade thresholds.

Reportedly, it achieves around 90 percent accuracy in flagging suspicious activity, processing large volumes of on-chain data more flexibly than traditional rule-based systems that rely on static heuristics.

2. Why This Matters For Crypto Users

China has banned most public crypto trading, yet its Supreme Peoples Procuratorate recorded 3,259 money-laundering prosecutions involving virtual assets and underground finance in 2025, showing crypto is still used in illicit flows.

This AI framework strengthens the states ability to trace those flows across chains and services, increasing risk for intermediaries that touch Chinese-linked activity, including OTC brokers, informal payment networks, and cross-border stablecoin routes.

Western agencies like the FBI and Europol are also rolling out AI-based tracing, but Chinas approach appears tightly integrated with national security infrastructure, which could lead to faster, coordinated crackdowns when patterns are detected.

3. Limits, Risks, And What To Watch Next

A 90 percent hit rate still implies false positives and misses, so human investigators must review AI-generated alerts; misuse could lead to over-blocking or wrongful suspicion, especially in complex DeFi flows.

Criminals can also weaponize open-weight AI models to optimize laundering routes, creating an arms race where both enforcement and attackers iterate rapidly on tactics.

Key things to watch include: whether China shares this technology with other jurisdictions, how deeply it is integrated into cross-border controls, and whether similar AI frameworks become standard in compliance tools used by exchanges and custodians worldwide.

Conclusion

Chinas move to deploy AI for crypto laundering detection signals a shift from simple bans toward sophisticated surveillance and forensic enforcement.

For the broader market, it reinforces a global trend: as AI makes both tracing and evasion more powerful, the edge increasingly goes to participants who understand how their flows look under algorithmic scrutiny and adjust toward cleaner, well-documented transaction patterns.