Guidance · HKMA · 15 Oct 2026
Authorized Institutions must develop overall A.I. strategies, designate an “owner” for each deployed A.I. use case, and tailor management of the full A.I. adoption lifecycle according to assessed risks and materiality. These institutions must apply the Guidance proportionately, drawing on existing governance and risk management arrangements. Approaches to human oversight must evolve with the A.I. landscape, moving from “human-in-the-loop” to “human-on-the-loop” or “human-in-command” arrangements depending on the nature and risks of the concerned use case.
55th Floor, Two International Finance Centre, 香 港 中 環 金 融 街 8 號 國 際 金 融 中 心 2 期 55 樓 8 Finance Street, Central, Hong Kong 網 址:www.hkma.gov.hk Website: www.hkma.gov.hk Our Ref.: B1/15C 15 October 2026 The Chief Executive All Authorized Institutions Dear Sir/Madam, Responsible Adoption of Artificial Intelligence I am writing to draw your attention to the Hong Kong Monetary Authority (HKMA)’s latest Framework for Responsible Adoption of Artificial Intelligence 1 (the Framework). The Framework supports Authorized Institutions (AIs) in harnessing the opportunities offered by artificial intelligence (A.I.) in a responsible manner. This marks important progress under the “Fintech 2030” strategy and Fintech Promotion Blueprint 2 to promote an A.I.-enabled banking ecosystem with Accountability, Resilience and Trust (collectively as “ART”). With A.I. adoption reaching a more mature state, and drawing on practical experience from the Generative Artificial Intelligence (GenA.I.) Sandbox 3 , as well as the latest guidance issued by international bodies including the Financial Stability Board 4 , the HKMA has updated its supervisory guidance for responsible adoption of A.I. (the Guidance, Annex A). The Guidance is anchored around the three pillars of the ART Framework and focuses on how AIs can demonstrate effective governance and accountability over A.I. adoption, more effectively manage the unique risks posed by A.I. in a proportionate manner, and build trust and transparency around A.I. solutions. /…Page 2 1 HKMA Press Release, “HKMA announces ‘ART’ Framework to promote responsible A.I. adoption”, issued on 15 October 2026 2 HKMA Press Release, “HKMA unveils Fintech Promotion Blueprint: From adoption to advancement”, issued on 3 February 2026 3 HKMA Press Release, “HKMA announces second cohort of GenA.I. Sandbox to advance responsible A.I. innovation”, issued on 15 October 2025 4 Financial Stability Board Report, “Sound Practices for Responsible Adoption of Artificial Intelligence (AI)”, issued on 15 October 2026
2 -
In particular, AIs are expected to develop overall A.I. strategies, designate an “owner” for each A.I. use case deployed, and also tailor their management of the full A.I. adoption lifecycle according to the assessed risks and materiality of a use case. The updated principles also acknowledge that approaches to human oversight must evolve with the A.I. landscape, including to move from “humanin-the-loop” to “human-on-the-loop” or “human-in-command” arrangements, depending on the nature and risks of the concerned use case. AIs are expected to apply the Guidance proportionately to their A.I. adoption and manage associated risks, drawing on existing governance and risk management arrangements as appropriate. To complement the Guidance, the HKMA also publishes a report on “Practical Insights from the Second Cohort of the GenA.I. Sandbox” (Practical Insights Report, Annex B). The second cohort of the GenA.I. Sandbox, which was officially concluded at the “Finance+A.I. Forum” today, has generated practical insights for the adoption of GenA.I. in banking. There is growing industry interest in agentic A.I., including systems that can assist in planning and executing multi-step tasks. These developments present new opportunities for AIs to enhance risk management, anti-fraud efforts and customer experience. They also underscore the need for sound governance arrangements as A.I. systems become more capable and autonomous. The Practical Insights Report summarises observations and good practices from the Sandbox pilots and includes cross-references to the Guidance to assist AIs in implementing the supervisory principles. AIs are encouraged to review the report and consider the relevance of observed good practices to an institution’s A.I. initiatives. Thirdly, the HKMA also issues a Concept Note on the Governance of Agentic A.I. (Concept Note, Annex C). The Concept Note sets out preliminary considerations for governance arrangements as agentic A.I. develops further in the banking sector. It will inform future formulation of supervisory requirements, drawing on experience from the GenA.I. Sandbox++ 5 and engagement with the industry. The HKMA will shortly issue a detailed discussion paper to seek views from AIs and relevant stakeholders. /…Page 3 5 HKMA Press Release, “First cohort of GenA.I. Sandbox++”, issued on 27 August 2026
3 -
The HKMA will continue to monitor developments in A.I. and collaborate with the industry to facilitate responsible innovation. Should you have any questions regarding this circular, please contact us at GenAI_sandbox@hkma.gov.hk. Yours faithfully, Carmen Chu Executive Director (Banking Supervision) Encl.
Note from RegAlert. AI assistants can read this document in full, and search 70,000+ more, through the RegAlert MCP connector (https://mcp.regalert.today/mcp). Free with an account. How to connect ChatGPT, Claude or Cursor.
More like this from HKMA
HKMA published 7 documents in the last 30 days. We email you each new one the day it's published.
Source: Hong Kong Monetary Authority — original document · Summary generated with machine assistance and reviewed before publication; the authoritative text is the regulator's original document. How RegAlert works
In RegAlert: added · updated