AIWiki
Malaysia
Back to all articles
Ethics & PolicyAI governanceOECDAI policy

OECD AI Principles

4 min readUpdated October 2026
OECD AI Principles
Formal name
Recommendation of the Council on Artificial Intelligence
Instrument
OECD/LEGAL/0449
Adopted
22 May 2019 (Ministerial Council Meeting)
Revised
8 November 2023 and 3 May 2024
Adherents
47 (38 OECD members, the European Union and non-member adherents including Singapore)
Legal force
Non-binding Recommendation (soft law)
The OECD AI Principles are the first intergovernmental standard on artificial intelligence. Formally the Recommendation of the Council on Artificial Intelligence (OECD/LEGAL/0449), they were adopted by the OECD Council at Ministerial level on 22 May 2019, revised on 8 November 2023 to update the definition of an "AI system", and substantively revised again at Ministerial level on 3 May 2024 to address generative and general-purpose AI.[1][2]

Structure

The instrument has two sections. Section 1 sets out five values-based principles for the responsible stewardship of trustworthy AI: inclusive growth, sustainable development and well-being; human rights and democratic values, including fairness and privacy; transparency and explainability; robustness, security and safety; and accountability. Section 2 contains five recommendations to policymakers: invest in AI research and development; foster an inclusive, AI-enabling ecosystem; shape an enabling, interoperable governance and policy environment; build human capacity and prepare for labour market transition; and international co-operation for trustworthy AI.[1][3]

Adoption and revision

The 2019 adoption made the OECD the first body to publish an AI-specific intergovernmental standard, developed with input from the G20; the G20 adopted compatible AI principles at the Osaka summit in June 2019.[2][3] The May 2024 revision, proposed by the OECD's Working Party on AI Governance, strengthened language on accountability, transparency and safety, and explicitly addressed risks across the lifecycle of general-purpose AI (GPAI) models, information integrity against disinformation, safety mechanisms such as override and decommissioning, privacy, intellectual property and environmental sustainability.[1][4]

As of the 2024 update the Principles have 47 adherents: the 38 OECD member countries, the European Union, and non-member adherents that include Argentina, Brazil, Egypt, Malta, Peru, Romania, Singapore and Ukraine.[1][5] Adherence is a voluntary political commitment — a Recommendation is, in the OECD's own classification, not legally binding and carries no sanction.[5]

Influence

The Principles function as the reference point for later, harder instruments. They shaped the EU AI Act, the G7 Hiroshima AI Process, ISO/IEC 42001 and the NIST AI Risk Management Framework, and are tracked through the OECD AI Policy Observatory (oecd.ai) and the OECD AI Incidents Monitor.[3][4] Many national AI strategies and ethics guidelines — including corporate governance programmes used by multinationals — cite the Principles as their baseline.[2]

>See Also

🇲🇾Malaysian Context

🇲🇾 Malaysia is not among the 47 adherents — Singapore is the only ASEAN member that adheres — yet the OECD Principles are explicitly written into Malaysian policy. The National Guidelines on AI Governance & Ethics (AIGE), issued by MOSTI on 20 September 2024, state that Malaysia's seven AI principles (fairness; reliability, safety and control; privacy and security; inclusiveness; transparency; accountability; and pursuit of human benefit and happiness) were formulated "in alignment with principles established by UNESCO, the OECD, the European Commission, and others".[6][7]

For Malaysia this alignment is deliberate strategy rather than membership: the AIGE documents state that Malaysia supports a risk-based approach precisely because alignment with international approaches supports national competitiveness and cross-border interoperability. Malaysian firms exporting to or supplying the EU, and Malaysian data centres and technology vendors serving global customers, are assessed against this shared vocabulary, while NAIO, MDEC and the Malaysia AI Governance Framework use it to anchor local guidance alongside the PDPA. Non-adherence also leaves Malaysia free to tailor implementation to local priorities such as Bahasa Malaysia support, multi-ethnic inclusion and Islamic finance compatibility.[6][7]

References

  1. ↑CASRAI. (2026). The OECD AI Principles: What 47 Adherents Actually Signed. https://casrai.org/guides/oecd-ai-principles-explained
  2. ↑AI Industry Innovation Alliance. (2025). OECD AI Principles. https://en.aipia.it/oecd-ai-principles
  3. ↑BIDDA. (2025). OECD Recommendation of the Council on Artificial Intelligence (2019, Updated May 2024). https://bidda.com/intelligence/oecd-ai-principles-2019-2024-update-trustworthy-ai
  4. ↑PACMap. (2026). OECD AI Principles (OECD/LEGAL/0449). https://pacmap.dev/regulation/oecd-ai-principles-2019
  5. ↑OECD AI Policy Observatory. OECD AI Principles. https://oecd.ai
  6. ↑MOSTI. (2024). National Guidelines on AI Governance & Ethics (AIGE). https://rmic.upsi.edu.my/wp-content/uploads/2025/03/NATIONAL-GUIDELINES-OF-AIGE.pdf
  7. ↑regulations.ai. (2024). National Guidelines on AI Governance & Ethics (AIGE). https://regulations.ai/regulations/RAI-MY-NA-NGAGEXX-2024
Back to all articles