Lever 2 · Updated 2026-05-03
⚖️ Lever 2 · Governance
Rules + Sandboxes + Law
Lever Definition
Give companies enough certainty to deploy
Take: "Permissive on training, strict on outputs" — Japan and Singapore are the only two countries doing this.
Projects
General Governance Frameworks (IMDA)
Model AI Governance Framework
IMDA · 2019
Asia's first AI governance framework
AI Verify + AI Verify Sandbox
IMDA · 10+ multinational enterprises participating
The world's first AI governance testing framework and toolkit
GenAI Eval Sandbox / GenAI Sandbox 2.0
IMDA · 2024–2025
Generative AI evaluation and experimentation environment
Generative AI Governance Framework
IMDA · 2024
Nine dimensions; one of the earliest dedicated GenAI governance frameworks globally
Agentic AI Governance Framework
IMDA · Released at Davos on 2026-01-22
The world's first Agentic AI governance framework
Trusted Data Sharing + DPTM Upgrade
IMDA · SS 714:2025
Trusted data sharing framework, with national standard upgraded
Financial Sector Governance 5-Layer Stack (MAS)
FEAT Principles
MAS · 2018
Four principles: Fairness / Ethics / Accountability / Transparency
Veritas Initiative v1 / v2 / v3
MAS · 2021–
Open-source toolkit for FEAT principles, pushing principles toward operability
Project MindForge
MAS · 24 institutions + Microsoft / AWS / Google / NVIDIA
Co-built GenAI risk framework for the financial industry
AI Risk Management Guidelines
MAS · 2024-12
Formal supervisory expectations document constraining bank use of AI
BuildFin.ai
MAS · Successor platform
Lets regulated financial institutions continuously test and report AI risk
Cybersecurity Governance (CSA)
Securing AI Systems Guidelines + Companion Guide
CSA · 2024-10
Full-lifecycle security guidelines for AI systems
Securing Agentic AI Addendum
CSA · 2025 (public consultation)
Security addendum focused on Agentic AI
Frontier AI Risk Advisory
CSA · 2025
Frontier AI risk advisory, coordinated with AISI
Cyber Trust Mark — AI Security Dimension
CSA · AI dimension extension
Cybersecurity certification extended to AI
Legal Governance (MINLAW + IPOS)
Copyright Act §244 — AI Training Exception
MINLAW · 2021
Tied with Japan as the world's most permissive copyright stance on AI training
IPOS "When Code Creates" Report
IPOS · 2024
Authorship stance on AI-generated content
Output-Side Strict-Control Quartet
MINLAW + MHA + MDDI · 2024–2025
OCHA + Elections Bill 2024 (deepfake ban) + Criminal Law Bill 2025 (criminalizing AI-generated intimate imagery) + Online Safety (Relief and Accountability) Bill 2025