OECD AI Principles

Applies To

Overview

Adopted by the Organisation for Economic Co-operation and Development (OECD) Council on 22 May 2019 and subsequently endorsed by G20 leaders in June 2019, the OECD AI Principles represent the foundational intergovernmental consensus on responsible AI development and deployment. The Principles emerged from extensive multi-stakeholder consultation and were developed by the OECD's Expert Group on AI (AIGO), drawing on input from governments, civil society, academia, and the private sector across 36+ member and partner nations. The framework is organized into two tiers. The first tier articulates five principles for trustworthy AI: (1) inclusive growth, sustainable development, and well-being; (2) respect for the rule of law, human rights, and democratic values, including fairness and privacy; (3) transparency and explainability; (4) robustness, security, and safety; and (5) accountability. The second tier sets out five recommendations directed at governments covering investment in AI R&D, fostering a digital ecosystem for AI, shaping an enabling policy environment, building human capacity, and pursuing international cooperation. The OECD AI Principles do not carry binding legal force in themselves but serve as a common reference model that has been directly incorporated or mirrored in numerous binding and non-binding instruments worldwide, including the EU AI Act and the G7 Hiroshima AI Process. The OECD revised the Principles in May 2024 to address generative AI, including safety, information integrity, and privacy. The OECD maintains a live AI Policy Observatory (oecd.ai) that tracks national implementation and policy developments against these principles. A significant operational development is the OECD Framework for the Classification of AI Systems, published as a companion document. It classifies AI systems along five dimensions: people and planet, economic context, data and input, AI model, and task and output. Regulators globally increasingly refer to this classification scheme when setting risk tiers.

Key Requirements

• •AI systems must be designed to be robust, secure, and safe throughout their lifecycle, with continuous risk assessment and mitigation.
• •Organizations deploying AI must ensure transparency and provide meaningful explanations of AI system outputs to affected stakeholders.
• •AI actors must be accountable for the proper functioning of AI systems and for respect of the other principles, based on their roles.
• •AI systems must respect the rule of law, human rights, democratic values, and diversity, and must include appropriate safeguards against misuse.
• •AI must be designed to foster inclusive growth and sustainable development and must not exacerbate inequality.
• •Governments are recommended to invest in AI research and development that addresses societal challenges.
• •Governments are recommended to develop agile policy and regulatory frameworks that respond to evolving AI capabilities.
• •International cooperation among governments and across stakeholders is explicitly recommended to share knowledge and develop compatible standards.

What Your Organization Must Do

• →Map your AI governance policies against the five OECD principles, noting they are voluntary, not legal duties.
• →Designate a named accountability owner for each material AI system and record it in a governance register.
• →Use the OECD classification dimensions to structure internal risk assessments of AI systems.
• →Document how you explain AI outputs to customers, employees and other affected stakeholders.
• →Track EU AI Act obligations separately, since that law is binding and draws on these principles.
• →Review the May 2024 revisions covering generative AI, safety, information integrity and privacy.

Playbook Guidance

Step-by-step implementation guidance for compliance teams.

Governance Controls

Operational controls that implement requirements from this regulation.

Frequently Asked Questions

Are the OECD AI Principles legally binding on companies?

No, the OECD AI Principles are a non-binding intergovernmental framework. They still carry practical weight because binding laws such as the EU AI Act, and voluntary codes such as the G7 Hiroshima code, build on them.

Which binding regulations are directly based on the OECD AI Principles?

The EU AI Act draws on them, including its definition of an AI system. The G7 Hiroshima AI Process code builds on them too, but it is voluntary. US Executive Order 14110 referenced them before it was revoked in January 2025.

Does the OECD AI Classification Framework create risk tiers that regulators actually use?

It does not set risk tiers itself. It classifies AI systems along five dimensions: people and planet, economic context, data and input, AI model, and task and output. Regulators and companies use it as a starting point for risk assessment.

Do the OECD AI Principles apply to private sector companies or only to governments?

Both. The framework has two tiers: five principles that apply to all AI actors including private sector developers and deployers, and five recommendations directed specifically at governments. Enterprises in financial services, healthcare, and critical infrastructure are explicitly within scope as AI actors.

What does accountability mean under the OECD AI Principles for an enterprise deploying AI?

AI actors are expected to be accountable for their AI systems functioning in accordance with all five principles, based on their role in the value chain. In practice this means designating a named accountability owner for each material AI system and documenting their responsibilities in a governance register.

How do the OECD AI Principles compare to the UNESCO Recommendation on the Ethics of AI?

Both are non-binding intergovernmental frameworks, but they differ in scope. The OECD Principles focus on trustworthy AI in economic and democratic contexts, while the UNESCO Recommendation puts more weight on human dignity and environmental sustainability.