On 2 November 2026, the Hong Kong Monetary Authority launched Project SAGE to support the responsible development and adoption of agentic finance in Hong Kong. It also established an Agentic Finance Expert Group bringing together financial institutions, technology providers, professional services firms and academia.
The HKMA uses “agentic finance” for financial activities in which AI agents may act with a degree of autonomy. Project SAGE is an exploratory initiative intended to support practical applications while developing the surrounding governance. It does not itself create a new regulatory requirement or confer approval on a particular product.
Four areas of work
Project SAGE covers four linked areas.
First, the HKMA will support use-case exploration and dialogue about implementation challenges. The announcement identifies consumer payments and agentic commerce, business procurement and treasury management, and machine-to-machine payments for digital services as possible areas of exploration.
Second, it will examine agent identity and delegated authority. The HKMA refers to approaches for verifying an AI agent's identity, “Know Your Agent” practices and authentication mechanisms. The practical question is not simply whether an agent can initiate an instruction, but how a firm establishes who authorised it, the scope and duration of that authority, and the controls that apply when the agent acts.
Third, the project will facilitate discussion of interoperability standards and protocols. Financial activity involving multiple agents, providers and institutions will depend on compatible ways to exchange instructions and evidence without losing the meaning of permissions, limits and accountability.
Fourth, Project SAGE will consider legal and regulatory issues, including governance arrangements, liability and consumer protection. The announcement frames these as matters for exploration and exchange, not settled conclusions.
Governance questions for firms
The initiative is relevant to firms considering agent-enabled payment journeys, treasury processes or customer services. A proportionate assessment could start with the activity the agent is expected to perform, the person or organisation on whose behalf it acts, and the limits on that authority.
The operating model should also make clear how the firm authenticates the agent and principal, records instructions and decisions, monitors outcomes, manages conflicts or unexpected behaviour, and stops or overrides the agent. Supplier dependencies, data access, incident response and accountability remain relevant even when a third party supplies the underlying model or platform.
These are general governance considerations rather than requirements announced by Project SAGE. The controls needed for a specific use case depend on the service, the agent's role and autonomy, the affected customers, the firm's regulatory position and the applicable law.
Status and source
Project SAGE was launched on 2 November 2026. It is an exploratory HKMA initiative supported by an expert group. The announcement is not a final rule, guidance instrument or implementation deadline.
This article provides general information, not legal advice or a conclusion about any firm's AI governance or regulatory compliance.