Chan calls for AI tools to combat digital-asset crime
HKMA Speech

Chan calls for AI tools to combat digital-asset crime

Hong Kong Monetary Authority (HKMA) executive Raymond Chan called for technology-led controls on digital-asset crime. Citing Chainalysis's 2026 report, he said illicit addresses received over $150 billion in 2025, with stablecoins accounting for 84 percent.

Following funds across chains

At the third workshop on AI adoption for financial-crime monitoring, Chan described risks as banks develop tokenised bonds, funds, real-world assets, deposits and stablecoin services.

These include volatility, operational and cyber risks, legal uncertainty and crypto-market contagion.

Speed, global reach, pseudonymity and limited reversibility complicate recovery of stolen assets; knowledge gaps and social media enable scams.

The HKMA is exploring agentic AI to connect on-chain transfers with off-chain customer transactions and detect crypto-fiat layering.

Chan envisaged combining blockchain analytics, customer due diligence, scam intelligence, suspicious-wallet monitoring and rapid freezing.

Linking wallets to identities before funds dissipate remains the central challenge.

A shared surveillance network

Hong Kong's risk-based framework follows Financial Action Task Force (FATF) standards.

The HKMA prioritises investor protection and financial-crime controls for stablecoins, with surveillance becoming operational and decentralised identity tools under exploration and piloting.

The Virtual Asset Intelligence Taskforce connects Hong Kong Police, the Customs and Excise Department, the HKMA, the Securities and Futures Commission, licensed virtual asset service providers and stablecoin issuers.

Overseas collaboration and FATF work target regulatory gaps and arbitrage.

Identity remains the bottleneck

Chan's central distinction is between tracing transactions and identifying their controllers quickly enough to stop funds dissipating.

AI remains an avenue under exploration, not a complete defence.

The framework therefore combines technology, intelligence sharing and cross-border cooperation rather than relying on one tool.

Report an error