Every AI Policy Ever · open in the interactive register

Published (non-binding, current)Non-binding · Guidelines / framework

Model Artificial Intelligence Governance Framework (Second Edition)

Record SG-01

Issued by
Personal Data Protection Commission (PDPC) and Infocomm Media Development Authority (IMDA)
Jurisdiction
Singapore
Level · region
National · South & Southeast Asia
Published / adopted
2020-01-21
Language
English

Summary

Voluntary framework of Singapore's Personal Data Protection Commission for organisations deploying AI at scale, in any sector and regardless of technology or algorithm. It turns common AI ethics principles into practices in four areas: internal governance structures and measures; the level of human involvement in AI-augmented decisions, set with a matrix of the probability and severity of harm (human-in-the-loop, over-the-loop or out-of-the-loop); operations management, covering data, explainability, repeatability, robustness, traceability, reproducibility and auditability; and stakeholder interaction and communication.

Safety topics

human oversightaccountability & liabilitytransparency & labellingrisk-based classification

Status notes

The first edition was released on 23 Jan 2019 and the second on 21 Jan 2020, both at the World Economic Forum Annual Meeting in Davos. The second edition added industry examples, clarified the human-over-the-loop approach, added robustness, reproducibility and auditability measures, replaced the use-case annex with a separate Compendium of Use Cases and limited algorithm audits to a regulator's request. IMDA and the PDPC developed a companion Implementation and Self-Assessment Guide for Organisations (ISAGO) with the World Economic Forum. Under the Personal Data Protection Act 2012, the IMDA is designated as the PDPC.

Sources

Open the official source https://www.pdpc.gov.sg/-/media/Files/PDPC/PDF-Files/Resource-for-Organisation/AI/SGModelAIGovFramework2.pdf

Secondary source: https://www.imda.gov.sg/about-imda/emerging-technologies-and-research/artificial-intelligence

Local copy

Archived. Last attempt 2026-10-18 03:22:31 UTC

Official source (third-party copy, not published) archive/SG-01/2026-10-11/official.pdf
Retrieved 2026-10-11 14:02:17 UTC
SHA-256 f2d199c826207a259c1c67b8379c0e2b0028fecf142129df9bf16447abcfb2d4
Secondary source (third-party copy, not published) archive/SG-01/2026-10-18/secondary.html
Retrieved 2026-10-18 00:48:27 UTC
SHA-256 4daf3adb0242074d79189f1e76611dcc24209536849166b952adf972c6d831fb

Verification

A
The instrument's own official text, or an official register, was stored and confirms the title, date and status. checked 2026-10-18 · official-url-fetched
Recheck of 18 Oct 2026. Stored and read: the framework PDF on pdpc.gov.sg (copy of 11 Oct 2026, unchanged when the archiver checked it on 18 Oct 2026), which gives the title, the edition table (first edition 23 Jan 2019, second edition 21 Jan 2020, both released at the WEF Annual Meeting in Davos), the PDPC's role, the 2020 IMDA/PDPC copyright and the framework's voluntary character. The IMDA AI page stored as the secondary source (copy of 18 Oct 2026) holds only site navigation and a last-updated date in its text layer, so nothing rests on it. Version 1's description of the framework as being for traditional, non-generative AI, and its remark that the generative and agentic AI frameworks supplement it, rest on no stored file of this record and were removed from the summary and notes; the links to SG-04 and SG-08 stay in related.
Second check: Independent check by Claude Code (claude-sonnet-5-5), 2026-10-18: 7 field(s) supported.

History

  1. 2026-10-18notes: The first edition was released on 23 Jan 2019. The second edition was released o → The first edition was released on 23 Jan 2019 and the second on 21 Jan 2020, bot — version 1's remark that the framework remains the baseline for traditional AI and is supplemented by the 2024 and 2026 frameworks rests on no stored file of this record and was removed; the second edition's changes, ISAGO and the IMDA/PDPC designation added from the PDF2026-10-18-recheck-south-asia-1 · Claude Code (claude-opus-5-5)
  2. 2026-10-18summary: Voluntary framework for organisations deploying traditional (non-generative) AI. → Voluntary framework of Singapore's Personal Data Protection Commission for organ — rewritten from the stored PDF: version 1's 'traditional (non-generative) AI' is not in the 2020 text; the four areas, the probability-severity matrix, the three levels of human involvement and the operations measures are2026-10-18-recheck-south-asia-1 · Claude Code (claude-opus-5-5)

Related records

Singapore MGF for Generative AI Singapore · Published (non-binding, current)

Singapore MGF for Agentic AI Singapore · Published (non-binding, current)

AI Verify Singapore · Published (non-binding, current)

Citation

Personal Data Protection Commission (PDPC) and Infocomm Media Development Authority (IMDA). Model Artificial Intelligence Governance Framework (Second Edition). 2020-01-21. https://www.pdpc.gov.sg/-/media/Files/PDPC/PDF-Files/Resource-for-Organisation/AI/SGModelAIGovFramework2.pdf. Status: Published (non-binding, current); verified 2026-10-18 (level A). Local copy archive/SG-01/2026-10-11/official.pdf retrieved 2026-10-11 14:02:17 UTC, SHA-256 f2d199c826207a259c1c67b8379c0e2b0028fecf142129df9bf16447abcfb2d4. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record SG-01.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.