Type: url
Status: generated
Summary
IMDA sets out a Model AI Governance Framework for Agentic AI (v1.5), describing what an AI agent is made of — model, instructions, memory, planning and reasoning, tools, protocols, controls, logging and monitoring — and the risks that follow from each, then organising governance into four dimensions: assess and bound the risks upfront, make humans meaningfully accountable, implement technical controls and processes, and enable end-user responsibility. It distinguishes an agent’s action-space from its autonomy, catalogues systemic and multi-agent risks including agent sprawl, miscoordination, conflict and collusion, and argues for deterministic system-level controls in preference to prompt-layer safeguards. Case studies from government agencies and industry illustrate each dimension, among them GovTech’s phased rollout of agentic coding assistants and Tencent’s CodeBuddy permission model.