OECD AI Principles

OECD AI Principles are a set of intergovernmental standards for trustworthy artificial intelligence adopted by the Council of the Organisation for Economic Co-operation and Development (OECD) in May 2019. They were the first international principles on AI endorsed at the intergovernmental level, and they establish both values-based principles for the responsible stewardship of AI and recommendations for governments on AI policy.^[1][2] The Principles were updated in May 2024 to address general-purpose and generative AI, and as of that update are adhered to by 47 countries and jurisdictions, including all OECD members, the European Union, and several non-member adherents.^[3]

History and Background

The Principles were developed through two years of multi-stakeholder consultation involving governments, industry, civil society, and technical experts, and were adopted at the OECD Council's Ministerial level in May 2019.^[4] They were subsequently reflected in the G20 AI Principles adopted in June 2019, giving them an effective global reach beyond the OECD's membership.^[3]

The 2024 update revised the Principles in light of rapid technological developments, notably the emergence of large-scale general-purpose and generative AI systems such as ChatGPT. The update reaffirmed the original values while adding guidance on risk-based approaches, the lifecycle of AI systems, and the responsibilities of organisations that develop or deploy foundation models.^[1][3]

The Principles

The instrument consists of five values-based principles for trustworthy AI. The first is inclusive growth, sustainable development, and well-being, requiring AI to contribute to broad social progress. The second is human rights, democratic values, and fairness, including privacy protection and the prevention of bias and discrimination. The third is transparency and explainability, requiring meaningful disclosure of AI capabilities and limitations. The fourth is robustness, security, and safety, demanding that AI systems function reliably and withstand misuse. The fifth is accountability, requiring organisations to take responsibility for the proper functioning of AI systems.^[3][4]

Alongside the values are five recommendations directed at governments: 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 pursue international co-operation for trustworthy AI.^[3]

Influence and Implementation

The OECD AI Principles are widely regarded as the foundation of subsequent AI governance frameworks, influencing the EU AI Act, the United States NIST AI Risk Management Framework, and the ISO/IEC 42001 AI management standard, all of which echo the core themes of transparency, accountability, safety, and respect for human rights.^[4] The OECD.AI Policy Observatory maintains a live database of more than 1,000 national AI policy initiatives and provides tools such as the AI incident tracker and the Classification of AI systems, which countries use to benchmark their own policies.^[2][9]

The OECD also hosts the Global Partnership on Artificial Intelligence (GPAI), an initiative that promotes the responsible development of AI and works with member countries and non-member partners.^[1]

>See Also

• EU AI Act
• AI Regulation Malaysia
• Malaysia AI Governance Framework
• ASEAN AI Governance
• Responsible AI

References