What happened
On 2 November 2026, coinciding with Hong Kong FinTech Week, the HKMA launched 'Project SAGE' (Securing Agentic Finance with Governance and Enablement) and established a cross-sectoral Agentic Finance Expert Group of 22 members (banks, payment firms, cloud providers, professional firms, academia). Simultaneously the HKMA published a Discussion Paper on the use of Agentic A.I. by banks, seeking industry feedback on supervisory considerations for governing AI agents that can plan and act with discretion. The paper develops 'intent-anchored accountability' (a named person inside the institution remains answerable for the business intent an agent may pursue), asks banks to flag which material agents a supervisor should see, to set risk appetite in an 'agent harness' of permissions/tool lists/deterministic checks, and to rehearse a stop (kill-switch) mechanism. Responses are due by 1 February 2027. It builds on the HKMA's October 2026 supervisory guidance and the FSB's 15 October 2026 sound-practice report.
Why it matters
Hong Kong is one of the first major financial regulators to open a formal consultation specifically on agentic AI in banking. It sets supervisory expectations for agent autonomy, identity verification ('Know Your Agent'), liability allocation, and deployer accountability. For any bank or financial institution operating in Hong Kong (and cross-border groups with HK exposure), this signals that agentic-AI governance will be supervised and that intent-anchored accountability, agent inventory and kill-switch controls are becoming expected.
Action needed
HK-licensed (and globally operating) banks should review the discussion paper, nominate owners/accountable persons for each material AI agent, prepare a response to the consultation before 1 February 2027, and assess readiness against the October supervisory guidance on lifecycle governance and use-case inventories.