HKMA explores agentic AI to detect crypto money‑laundering, focusing on stablecoins
Hong Kong’s monetary authority is testing autonomous AI agents that combine blockchain analytics with bank data to spot illicit crypto activity, especially in stablecoins.
Key takeaways
- The HKMA is developing agentic AI systems that monitor both blockchain transfers and traditional bank transactions.
- Stablecoins represented 84% of illicit crypto transaction volume in 2025, according to Chainalysis.
- Hong Kong plans to extend crypto licensing to dealers, custodians, advisers and asset managers before the end of 2026.
The Hong Kong Monetary Authority (HKMA) announced that it is piloting autonomous artificial‑intelligence agents to help banks identify cryptocurrency‑related money‑laundering. The AI would analyse blockchain activity together with conventional banking data, aiming to spot suspicious fund movements across fiat and digital assets, according to a regulatory‑technology workshop on Nov. 17.
Agentic AI design and governance
The HKMA’s initiative, dubbed Project SAGE (Securing Agentic Finance with Governance and Enablement), seeks to create AI agents that can plan and execute multiple monitoring tasks within predefined limits. The regulator is also drafting technical standards for authenticating agents, defining their authority, and clarifying liability and consumer‑protection issues. Over 60% of applications reviewed in the regulator’s generative‑AI sandbox in October involved such agentic technologies.
Focus on stablecoins and illicit flows
Stablecoins are a particular focus because they provide price stability and are often held in self‑custodied wallets, making them attractive for laundering. Chainalysis’ 2026 Crypto Crime Report found that stablecoins accounted for 84% of illicit crypto transaction volume in 2025, with sanctioned entities’ receipts jumping 694% year‑over‑year. Although illicit activity still represents less than 1% of total crypto volume, the rapid growth prompted the HKMA to integrate blockchain analytics, customer‑identification records and fraud intelligence into a single detection system.
Regulatory backdrop and industry response
The AI effort aligns with broader regulatory moves in Hong Kong. The city’s Stablecoins Ordinance, introduced in August 2025, led to the first licences for HSBC’s RedCoin and Anchorpoint Financial’s institutional stablecoin earlier this year. The HKMA also runs the Virtual Asset Intelligence Taskforce, a joint platform of regulators, law‑enforcement and licensed crypto firms that shares intelligence and coordinates rapid interventions. Financial Secretary Paul Chan confirmed that new licensing legislation will broaden the scope to include dealers, custodians, advisers and asset managers before the end of 2026.
What to watch next
Key developments to monitor include the rollout of the agentic AI pilot, any formal governance framework released by the HKMA, and the impact of the upcoming licensing legislation on crypto service providers. Observers will also watch how quickly banks can integrate AI‑driven monitoring into existing AML workflows and whether the approach reduces the share of illicit stablecoin transactions.
Frequently asked questions
What is agentic AI and how does it differ from traditional AML tools?
Agentic AI refers to autonomous software agents that can plan and execute multiple monitoring tasks across different data sources, such as blockchain ledgers and bank transaction records, whereas traditional tools typically analyse only one domain.
Why are stablecoins highlighted in Hong Kong's anti‑money‑laundering efforts?
Stablecoins made up 84% of illicit crypto transaction volume in 2025, according to Chainalysis, due to their price stability and ease of transfer through self‑custodied wallets.
Sources
Get stories like this in your inbox
The day’s biggest crypto stories and market moves in one short morning email. Free, no spam.