The OECD AI Principles: values-based principles for trustworthy AI
The OECD AI Principles are the first intergovernmental standard for AI. They provide five values-based principles and policy recommendations to guide trustworthy and human-centric AI development. The definition of an AI system in these principles is widely used in international frameworks.
Overview and adoption
The OECD AI Principles are the first intergovernmental standard for AI. They promote AI that is innovative, trustworthy, and respectful of human rights and democratic values. the OECD AI Principles overview
The Principles define an AI system as a machine-based system that infers from input how to generate outputs such as predictions, content, recommendations, or decisions that can affect physical or virtual environments. This definition is now used by the European Union, the Council of Europe, the United States, and the United Nations in their legislative and regulatory frameworks.
There are 47 adherents to the Principles, including OECD members and non-members. The G20 AI Principles are drawn from the OECD AI Principles.
Values-based principles
The Principles outline five core values: supporting inclusive growth and well-being, respecting human rights and democratic values like fairness and privacy, ensuring transparency and explainability, guaranteeing security, safety, and reliability, and requiring accountability.
These values guide AI actors in the responsible development and use of human-centered AI systems. They also address ethical concerns such as privacy, autonomy, fairness, and explainability, as discussed in the Stanford Encyclopedia of Philosophy's entry on AI ethics. the Stanford Encyclopedia entry on AI ethics
The values emphasize that AI systems should be transparent and explainable to users, and that organizations must be accountable for the outcomes of their AI systems.
Recommendations for policymakers
The OECD AI Principles also include recommendations for policymakers to invest in AI research and development, foster an inclusive AI-enabling ecosystem, shape an interoperable governance and policy environment, build human capacity for labour market transitions, and pursue international cooperation for trustworthy AI.
These recommendations are designed to help governments create policies that promote innovation while managing risks. They aim to build a foundation for global interoperability between jurisdictions.
Global influence and practice
The OECD AI Principles have influenced other international initiatives. The Hiroshima AI Process, launched after the G7 Hiroshima Summit, developed a comprehensive policy framework with guiding principles and a code of conduct for advanced AI systems, building on the OECD's approach. the Hiroshima AI Process page
The United Nations High-level Advisory Body on AI similarly proposed a blueprint for global AI governance that emphasizes international cooperation and the protection of human rights. the UN AI Advisory Body
To support implementation, the OECD provides a Catalogue of Tools and Metrics for Trustworthy AI. This catalogue offers technical, educational, and procedural tools that help AI actors build systems that are fair, transparent, explainable, secure, and safe. It also allows users to contribute their own tools. the OECD Catalogue of Tools and Metrics
The catalogue can be browsed by tool type, origin, and scope. The flow of using the catalogue is shown in the figure.
| Principle | Focus |
|---|---|
| Inclusive growth, sustainable development and well-being | Ensures AI benefits all people and supports sustainable development |
| Human rights and democratic values, including fairness and privacy | Protects fundamental rights and avoids discrimination |
| Transparency and explainability | Users can understand how AI systems make decisions |
| Reliability, security and safety | AI systems operate dependably and resist misuse |
| Accountability | Organizations take responsibility for AI outcomes |
Frequently asked questions
What are the OECD AI Principles?
They are the first intergovernmental standard for AI, promoting innovative and trustworthy AI.
Which countries adhere to the OECD AI Principles?
As of the OECD's overview, 47 adherents, including OECD members and non-members such as Argentina, Brazil, and Singapore.
How do the G20 AI Principles relate to the OECD AI Principles?
The G20 AI Principles are drawn from the OECD AI Principles.
What definition of an AI system do the OECD AI Principles provide?
An AI system is a machine-based system that infers from input how to generate outputs like predictions, content, or recommendations.
How can the OECD Catalogue help with trustworthy AI?
It provides technical, educational, and procedural tools for developing AI systems that are fair, transparent, and safe.