Hong Kong puts AI-agent identity and accountability on its finance agenda

Hong Kong’s monetary authority launched Project SAGE on November 2, 2026, setting out work on how artificial intelligence agents could participate in financial services and how their actions should be governed. The official announcement describes an initiative to develop and test applications alongside safeguards, rather than a completed rulebook.

SAGE stands for Securing Agentic Finance with Governance and Enablement. Its launch brings together two questions often discussed separately: what an automated system can do, and what it has permission to do for someone else.

Identity is only part of the problem

The HKMA identified four areas of work: practical applications, agent identity, interoperability, and legal and regulatory arrangements. Potential applications include consumer payments, business procurement and payments between machines.

Its identity work will examine how to authenticate an agent and establish the authority delegated to it. In practical terms, knowing which software submitted an instruction is different from knowing whether a customer authorized that instruction. That distinction explains why authentication alone cannot resolve questions of responsibility.

The authority also established an expert group spanning financial institutions, technology providers, professional services and academia.

Governance before broad deployment

RTHK’s coverage reported that the HKMA is issuing a discussion paper on banks’ use of agentic AI, including responsibility arrangements and a possible register of trusted agents.

Those topics remain proposals for examination. A possible register should not be treated as an operational certification system, and the launch does not establish that an AI agent can safely handle every financial task. For readers assessing future services, the useful questions concern authorization, accountability and how errors can be challenged.

Production note: This original brief was researched and written with AI assistance for The Daily Blend Editorial, using the linked sources.

What to watch

• The discussion-paper process will provide the next public detail on proposed governance arrangements.
• Project SAGE’s expert group will support further work on use cases, identity and interoperability; the launch statement gives no completion date.