China trains AI to detect crypto laundering

TLDR

China is developing AI and large language model tools to detect crypto money laundering and other economic crimes involving virtual assets.

1. Researchers at Chinas National Police University built an AI-powered framework that reportedly flags suspicious crypto transactions with around 90% accuracy.
2. The system strengthens Chinas ability to enforce its strict crypto bans and monitor cross-border flows, increasing risk for illicit use of Bitcoin and other coins.
3. Globally, this signals an AI arms race in blockchain forensics, with regulators and criminals both adapting; effectiveness and false positives will be key to watch.

Deep Dive

1. How Chinas AI System Works

According to a detailed report on an AI-powered detection framework, researchers at Chinas National Police University, under the Ministry of Public Security, have built a system that combines pattern-recognition AI with large language models (LLMs) to track money laundering involving virtual assets.

The tool ingests large volumes of on-chain data, modelling transaction flows to spot classic laundering behaviors such as layering (multiple hops to obscure origin) and structuring (splitting amounts). LLMs help interpret unstructured context, like transaction memos or communication patterns, rather than relying only on rigid rules.

Tests cited in the report claim roughly 90% accuracy in flagging suspicious activity, a significant jump over traditional rule-based monitoring, although it still produces false positives that require human review.

2. Why This Matters For Crypto Users

China already bans most public cryptocurrency trading, yet its Supreme Peoples Procuratorate reported 3,259 money laundering prosecutions involving virtual assets and underground finance in 2025, underscoring ongoing illicit use despite restrictions.

An AI system that can automatically surface suspicious Bitcoin or stablecoin flows gives authorities a more powerful way to enforce capital controls and criminal law, especially when paired with plans by the forex regulator SAFE to crack down heavily on illegal cross-border financial activities using technologies like AI.

For legitimate users, the main risk is increased scrutiny if funds touch Chinese counterparties, OTC desks, or platforms that cooperate with Chinese investigations. For illicit actors, mixers, cross-chain hops and complex routing become less reliable obfuscation when AI can trace behavioral patterns across many wallets.

3. Global Enforcement And What To Watch

The same report notes that Western agencies such as the FBI and Europol are also deploying AI-based tracing tools, but Chinas framework appears tightly integrated with national security and economic crime infrastructure, potentially allowing faster operational use.

Key things to watch are:

1. Whether China shares or licenses this technology to other states or international bodies.
2. How often flagged flows lead to successful prosecutions versus false alarms.
3. Whether privacy tools, layer 2s, or new chain designs evolve specifically to defeat AI-based tracing.

If AI-driven forensics become standard, the cost of hiding large-scale laundering rises, while compliance and high-quality data become more valuable for exchanges and service providers.

Conclusion

Chinas move to train AI against crypto laundering fits a broader trend of regulators using advanced analytics to narrow the gap between pseudonymous blockchains and traditional financial surveillance. For crypto users, the practical impact is tighter monitoring of cross-border and high-risk flows, and a likely long-term split between transparent, regulated liquidity and harder-to-access grey markets, as both enforcement and evasion adapt to the same AI tools.