Jump to content

Hong Kong responsible AI supervisory guidance (2026)

From AI Law Wiki

AI-generated text. This page was generated using artificial intelligence.

Law & policy information
Name
Supervisory Guidance for Responsible Adoption of Artificial Intelligence[1]
Jurisdiction
Hong Kong; Authorized Institutions[2]
Instrument type
Supervisory guidance[1]
Bill / legal citation
HKMA circular B1/15C and Annex A[2]
Legislative or adoption status
Issued October 15, 2026[2]
Effective dates
Timely uplift of controls; no fixed implementation deadline specified in Annex A[1]
Last source verification
October 15, 2026; full circular, Annex A and Annex C reviewed[2][1][3]

Overview

The Hong Kong Monetary Authority issued its ART framework and updated supervisory guidance on October 15, 2026 for Authorized Institutions. ART stands for Accountability, Resilience and Trust. The circular links the framework to Fintech 2030, sandbox experience and the Financial Stability Board’s work.[2]

Scope and supervisory expectations

Annex A supersedes the November 2019 high-level AI principles. Institutions should strengthen controls promptly and proportionately, using existing governance arrangements where appropriate.[1]

The guidance addresses board-approved strategy, accountable use-case owners, approved-use inventories and human oversight. It calls for risk-based testing, monitoring, recovery and audit records. Higher-risk uses warrant stronger validation and emergency suspension controls. Institutions retain responsibility for third-party AI, with data, contractual and cybersecurity safeguards. Explainability, fair outcomes and routes for human review support customer protection.[1]

Agentic AI: preliminary concepts

The separate concept note explores connecting agents’ actions to business intent, identifying accountable business functions, mapping dependencies and setting operating boundaries. It considers an industry-level registry for material agents and human intervention before consequential actions. These are preliminary considerations for discussion; the note expressly says they are not specific supervisory requirements at this stage. A detailed discussion paper will seek industry views.[3]

Implementation

The circular expects institutions to assign owners and adapt lifecycle controls to risk. It encourages consideration of a separate sandbox practical-insights report. The concept note is intended to inform future supervisory requirements after experimentation and industry engagement.[2]

References

  1. ↑ 1.0 1.1 1.2 1.3 1.4 1.5 HKMA, Annex A: Supervisory Guidance for Responsible Adoption of Artificial Intelligence, pp.1–8.
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 HKMA, Responsible Adoption of Artificial Intelligence, circular B1/15C, October 15, 2026, pp.1–3.
  3. ↑ 3.0 3.1 HKMA, Annex C: Concept Note on Governance of Agentic Artificial Intelligence, pp.1–2.
Retrieved from "https://ailawwiki.com/index.php?title=Hong_Kong_responsible_AI_supervisory_guidance_(2026)&oldid=6604"