Singapore governs enterprise AI through 27 instruments and a risk-based stance: voluntary frameworks, binding adjacent law, and regulation held in reserve

Asked (summary):

What are the published policies and parliamentary discussions on AI in Singapore? For each discussion, highlight the key features relevant to a decision-making enterprise framework.

As of 12 October 2026 the research identified 27 published Singapore instruments with material relevance to AI strategy, governance, assurance, privacy, cybersecurity or enterprise implementation, and 14 substantive parliamentary discussions from 2023 onward — a researched selection strongest for 2026, not a claim to every historical mention of AI in Parliament. Singapore has no omnibus AI Act: voluntary and advisory frameworks sit alongside binding existing laws and sector rules, and the Government has said it may regulate where existing measures prove inadequate.

Where each instrument acts in the enterprise AI decision lifecycle

Each row is one published instrument, newest first; a mark means the instrument speaks to that lifecycle stage, and its colour states the instrument's legal status. Solid rows are the core AI-governance layer; hollow-dot rows are supporting data, cyber and sector controls. Hover or tap a row for its published features, enterprise relevance and source.

mandatory voluntary advisory proposed strategic status not stated square = core AI governance · ring = supporting data / cyber / sector control

Hover or tap a row above to read the instrument's published features, its enterprise relevance and its source.

Voluntary does not mean optional governance: the July 2026 answer on high-risk AI anchors any future regulation to deployment context, severity of harm and adequacy of existing law — employers already face the Workplace Fairness Act, MAS and MOH sector guidance. telescope.gov.sg

The Model AI Governance Framework for Agentic AI (updated May 2026, 8 corroborating pages) expects designated oversight roles, human approval checkpoints and controls proportionate to an agent's autonomy — in Parliament the Government confirmed it stays voluntary with no binding agentic rules in the near term. imda.gov.sg

Testing is a first-class requirement: AI Verify (11 governance principles, crosswalked to the NIST AI RMF), Project Moonshot LLM red-teaming, the GenAI evaluation sandbox and the Global AI Assurance Sandbox give enterprises a pre-deployment assurance stack. imda.gov.sg

October 2026 in Parliament: the onus is on organisations — including those outside critical information infrastructure — to safeguard themselves against rogue AI-agent attacks, since agents and bad actors exploit the same unlocked doors. straitstimes.com

14 parliamentary discussions, mapped to enterprise-control domains

Sitting date, the enterprise-control domain each discussion bears on, the published government position and the decision implication for deployers.

All 27 instruments

Legal status as published; blank cells were blank in the source and are left blank, not inferred. Corr. = corroborating pages; rows with 1 are source-linked, not independently corroborated.

Instrument	Updated	Status	Enterprise relevance	Corr.	Source

Method: 27 Singapore policy instruments and 14 parliamentary discussions (2023–2026) researched to 12 October 2026. Policy discovery scanned 970 URL-deduplicated web results across 10 searches; parliamentary discovery scanned 780 across 10 searches. Sources span Singapore government and agency sites, Parliament-derived records and independent news. Each cell reports published text; a mark in the lifecycle matrix is an editorial mapping of an instrument's published scope to a stage. Features and relevance text are trimmed to about 420 characters for space; most rows rest on one primary page unless corroborating pages exceed 1. Null cells remain blank.

This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors. Nothing in this report is investment, legal, financial, or other professional advice. All trademarks and referenced content remain the property of their respective owners; no affiliation or endorsement is implied. To report an error, rights concern, or request removal: legal@keenable.ai.