OECD AI Principles for Trustworthy Artificial Intelligence
What the 2019 OECD AI Principles ask of organizations and governments, and how the OECD framework for classifying AI systems helps you assess an AI tool.

A company adds an AI screening tool to its hiring process. A rejected applicant asks why. If nobody can explain the decision, and nobody owns the answer, the company has a problem that no software update will fix.
That gap, between using AI and being accountable for it, is what the Organisation for Economic Co-operation and Development (OECD) set out to close. Its AI Principles, adopted in May 2019, were the first intergovernmental standard on AI, and they remain a sound checklist for any organization deciding how to use AI responsibly.
OECD.AI Policy Observatory
The OECD AI Policy Observatory (OECD.AI) is a hub for resources across the OECD and its partners. It aims to foster dialogue and supply multidisciplinary policy analysis and data on AI’s areas of impact. Its country dashboards track AI policy initiatives from dozens of countries and the European Union.
For a closer look at the OECD.AI Policy Observatory, you can explore it here.
Towards a Human-Centric Approach to Trustworthy AI
The OECD Principles on Artificial Intelligence promote AI that is innovative, trustworthy, and respectful of human rights and democratic values. The OECD Council adopted them in May 2019 as a Recommendation, and the G20 drew on them for its own AI principles the following month.
Further details on the OECD Council Recommendation on Artificial Intelligence
Understanding the OECD AI Principles
The principles contain five values-based principles for responsible stewardship of trustworthy AI and five recommendations to governments.
Principles for Responsible Stewardship of Trustworthy AI
- Inclusive Growth, Sustainable Development, and Well-being: AI should augment human capabilities, enhance creativity, advance the inclusion of underrepresented populations, reduce inequalities, and protect natural environments.
- Human-centered Values and Fairness: AI systems should respect the rule of law, human rights, democratic values, and diversity. They should also supply safeguards appropriate to the context, such as a capacity for human determination.
- Transparency and Explainability: AI systems should foster a general understanding and awareness of their interactions. Affected individuals should be able to understand and challenge the outcomes of an AI system.
- Robustness, Security, and Safety: AI systems should maintain robustness, security, and safety throughout their entire lifecycle. It is crucial to continuously evaluate and manage potential privacy, digital security, safety, and bias risks.
- Accountability: AI actors should be accountable for the proper functioning of AI systems in line with the above principles.
National Policies and International Cooperation for Trustworthy AI
- Investing in AI Research and Development: Governments should encourage public and private investment in research and development to spur innovation in trustworthy AI.
- Fostering a Digital Ecosystem for AI: Governments should ease the development of a digital ecosystem that supports safe, fair, legal, and ethical data sharing.
- Shaping an Enabling Policy Environment for AI: Governments should promote a policy environment that supports an agile transition from the research and development stage to the deployment and operation stage for trustworthy AI systems.
- Building Human Capacity and Preparing for Labor Market Transformation: Governments should collaborate closely with stakeholders to prepare to transform the world of work and society. They should also supply support for workers affected by AI deployment.
- International Cooperation for Trustworthy AI: Governments should cooperate across borders and sectors to advance responsible stewardship of trustworthy AI.
The OECD Framework for Classifying AI Systems
The OECD framework for classifying AI systems offers a multifaceted approach to evaluating AI technologies, covering five key dimensions: People & Planet, Economic Context, Data Input, AI Model, and Task & Output. Each dimension provides a unique perspective to assess AI’s implications, risks, and benefits, ensuring responsible, ethical, and effective development and use. This framework is crucial for organizations to understand AI systems’ societal and technical impact.
- People and Planet: Focuses on AI’s impact on human rights, the environment, and societal concerns, emphasizing privacy and ethical issues.
- Economic Context: Evaluates AI systems within their financial environment, considering aspects like industry sector, business model, operational criticality, deployment impact, scale, and technological maturity.
- Data Input: Examines data types, collection methods, structure, and the inclusion of expert input in AI models, emphasizing data quality and reliability.
- AI Model: Concentrates on the technical construction and application of AI models, including algorithm use, design, and development methodologies.
- Task and Output: Assesses AI systems based on the tasks they perform, their outputs, resultant actions, and the effectiveness of these tasks and outputs.
What This Means for Your Organization
The OECD aims to supply a platform for exchanging information on AI policy and activities, fostering multi-stakeholder and interdisciplinary dialogue to promote trust in and adoption of AI. For an organization, the practical test is simple: for each AI system you use, can you say who is accountable for it, explain its outcomes to the people it affects, and show how its security and safety risks are managed?
References
- OECD, Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449, adopted May 22, 2019. https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449
- OECD.AI Policy Observatory, OECD AI Principles overview. https://oecd.ai/en/ai-principles
- OECD, OECD Framework for the Classification of AI Systems, OECD Digital Economy Papers No. 323, February 2022. https://www.oecd.org/publications/oecd-framework-for-the-classification-of-ai-systems-cb6d9eca-en.htm
- OECD, The State of Implementation of the OECD AI Principles Four Years On, OECD Artificial Intelligence Papers No. 3, 2023. https://www.oecd.org/publications/the-state-of-implementation-of-the-oecd-ai-principles-four-years-on-835641c9-en.htm