Start by checking which law, supervisory expectation, proposal, or guidance applies to your organization. The table identifies each instrument, its status, and who it addresses. The contribution section then explains what AAES Govern records can help a reviewer examine.
These pages describe the product and cite public instruments. They are not legal advice, not a certification, and not a statement that any instrument applies to your organization. The official source is authoritative; applicability questions belong to your advisers.
Instruments, by status
| Instrument | Status | What it addresses |
|---|---|---|
| Model AI Governance Framework for Agentic AI · IMDA | Published, voluntary. Version 1.0 launched 22 January 2026; version 1.5 issued 20 May 2026 and updated 5 June 2026. Not legislation; creates no legal duty. | Any organization deploying AI agents. Four dimensions: assess and bound risk upfront, meaningful human accountability, technical controls and processes, end-user responsibility. |
| Guidelines on Artificial Intelligence Risk Management · MAS | Final supervisory guidelines; not legislation. Issued 7 October 2026 after the P017-2025 consultation (response to feedback). They take effect on 7 October 2027: Sections 3 to 4 apply from that date, Sections 5 and 6 by 7 October 2028. MAS intends to consult in 2027 on additional guidance for agentic AI. See the MAS announcement. | All financial institutions and all forms of AI, applied in proportion to risk materiality. Expectations cover board and senior-management oversight, AI inventories and risk-materiality assessment, lifecycle controls including human oversight, monitoring and change management, and third-party AI risk. MAS names agentic AI systems that operate autonomously and access tools as a reason to review controls regularly. |
| Personal Data Protection Act 2012 · PDPC | Binding law. | Personal data in AI systems, including agentic ones. Using AAES does not discharge the organization's accountability obligation; data intermediaries can carry their own duties. |
| Safeguards for Agentic Finance at Runtime (SAFR) · MAS and industry | Industry white paper, not binding law. Published 3 July 2026 under MAS' BuildFin.ai initiative. It proposes a framework for runtime safeguards; publication does not establish a legal duty or certify an implementation. | Agentic finance workflows. Governance checkpoints verify and record proposed actions before execution, with policy-bound execution, real-time validation, auditability and interoperability. AAES' routed authorization and decision records can be examined against selected themes; no SAFR conformance assessment of AAES has been published. See the MAS launch announcement. |
| FEAT Principles · MAS (2018) | Non-binding principles, referenced in MAS supervisory dialog with financial institutions. | Fairness, ethics, accountability and transparency in AI-driven decisions; action-layer records can support review of accountability and transparency. |
What AAES Govern records may contribute
| Expectation | AAES contribution | Capability status |
|---|---|---|
| Bounded agent authority (IMDA dimension 1; selected SAFR runtime-control themes) | Registered capabilities with risk tiers, scoped short-lived grants, and refusals recorded before execution, on actions routed through AAES Govern. | Implemented; demonstrated in the evaluation package. Enforcement depends on control of the agent's credential path. |
| Human accountability (IMDA dimension 2; MAS Guidelines oversight expectations) | Every registered agent has a named accountable manager whose chain of managers ends at a person. Actions registered as irreversible require approval by an authorized person; an agent cannot approve itself. | Implemented. |
| Technical controls and evidence (IMDA dimension 3; MAS Guidelines lifecycle controls; selected SAFR auditability themes) | Decision records are linked by cryptographic fingerprints to help detect changes. The client's own reviewer can check an export offline. | Implemented. An independently controlled witness can be configured separately; that is not an external assessment, and the default deployment has none. |
AAES Govern decision records may support review of access and approval controls. They do not establish lawful processing, model fairness, suitability, or compliance with any instrument as a whole.
Export checks do not establish complete capture, downstream success, or meaningful human oversight. Enforcement requires a credential path the agent cannot bypass. See the verification checks and their limits.
The manager assignment is a configured association, not a statutory appointment. The operator classifies an action as irreversible; this is not an independent assessment of its effects.
Eligible manager agents may approve other actions where policy permits, but cannot satisfy mandatory human approval. Policies can also require approval for other actions: the irreversible classification does not identify every action needing human oversight.
An approval record alone does not show that the approver had adequate information, time, or practical power to intervene.
Your responsibilities
- The client remains the regulated entity. Ask your advisers which provisions of the final MAS Guidelines apply to you and what they require.
- Activity outside AAES Govern is visible only where telemetry is supplied. Decision records cover routed requests; reported telemetry may describe other activity without establishing complete coverage.
- AAES has no independent certification or assessment. No SOC 2 report exists, and no penetration test has been performed yet; an independent test is planned for Q4 2026.
- Status changes. MAS has said it will consult in 2027 on additional guidance for agentic AI. Re-check the primary sources before relying on the status shown here.
Bring the instruments your risk team cites to a scoped evaluation.