AI Regulation ReferenceDoes Model AI Governance Framework require Human Oversight?
Singapore • voluntary
Yes — 1 provision
Requirements at a glance
This regulation imposes 4 specific requirements for Human Oversight across 1 provision:
- Human-in-the-loop — Appropriate level of human involvement based on risk and impact
- Decision models — Three models: human-in-the-loop, human-on-the-loop, human-out-of-the-loop
- Risk-proportionate — Level of oversight proportionate to risk of AI application
- Agentic AI oversight — 2026 update adds guidance for autonomous agent monitoring and intervention
AI Governance and Human Oversight #
Requirements
| Requirement | Details |
|---|---|
| Human-in-the-loop | Appropriate level of human involvement based on risk and impact |
| Decision models | Three models: human-in-the-loop, human-on-the-loop, human-out-of-the-loop |
| Risk-proportionate | Level of oversight proportionate to risk of AI application |
| Agentic AI oversight | 2026 update adds guidance for autonomous agent monitoring and intervention |
Penalties
| Violation | Fine |
|---|---|
| Non-compliance | Voluntary — no direct penalties; organizations legally accountable for AI actions under existing law |