Governance as strategy, not checklist: Why agentic governance must rise with autonomy. A feedback submission on the Model AI Governance Framework for Agentic AI (v1.5)

This paper is a feedback submission to IMDA on the Model AI Governance Framework for Agentic AI (Version 1.5). While the framework explicitly takes into account that governance needs to be devised upfront for agentic ai and outlines four principles: assessing risks upfront, making humans meaningfully accountable, implementing technical & process controls, and enabling end-user responsibility - designing the agents with a matrix on action-space and autonomy axes, when applied through a lens of GRC (governance, risk, and compliance), the framework’s top-level human oversight can be setup as an exception-based observation for the most autonomously acting multi-agent systems carrying the greatest business risk, receiving the non-stringent governance (as indicated in the MSD case study for calibrating governance processes to level of agency). The submission argues for treating governance as strategy: a board-owned capability that rises with autonomy even as human involvement in each action falls.