OECD AI Principles

Jurisdiction:

OECD

voluntary

Effective:

May 22, 2019

Authority:

OECD Council

Official text Verified Mar 26, 2026

Obligations Covered

Transparency & Disclosure Bias & Discrimination Prevention Risk Assessment Human Oversight

Transparency and Explainability (Principle 1.3) #

Obligation:

Transparency

voluntary

Effective:

May 22, 2019

Risk tier:

all

Scope:

providers, deployers

Requirements

Requirement	Details
Transparency commitment	AI actors should commit to transparency and responsible disclosure
Meaningful information	Provide information appropriate to the context to foster understanding
Explainability	Enable people affected by AI systems to understand and challenge outcomes
Awareness of AI interaction	People should be able to know when they are interacting with AI

Penalties

Violation	Fine
Non-compliance	Voluntary — no binding enforcement mechanism

Human-Centred Values and Fairness (Principle 1.2) #

Obligation:

Bias Prevention

voluntary

Effective:

May 22, 2019

Risk tier:

all

Scope:

providers, deployers

Requirements

Requirement	Details
Human rights respect	AI actors should respect rule of law, human rights, and democratic values
Fairness	Ensure AI does not produce unjust or discriminatory outcomes
Privacy and data protection	Respect privacy rights throughout the AI lifecycle
Non-discrimination	AI should not create or reinforce unfair bias

Penalties

Violation	Fine
Non-compliance	Voluntary — no binding enforcement mechanism

Robustness, Security and Safety (Principle 1.4) #

Obligation:

Risk Assessment

voluntary

Effective:

May 22, 2019

Risk tier:

all

Scope:

providers, deployers

Requirements

Requirement	Details
Risk management	AI systems should not pose unreasonable risks; manage risks throughout lifecycle
Robustness	Ensure AI systems function as intended under normal and adversarial conditions
Security	Address cybersecurity risks in AI systems
Traceability	Enable traceability of AI system outcomes to data and processes

Penalties

Violation	Fine
Non-compliance	Voluntary — no binding enforcement mechanism

Accountability (Principle 1.5) #

Obligation:

Human Oversight

voluntary

Effective:

May 22, 2019

Risk tier:

all

Scope:

providers, deployers

Requirements

Requirement	Details
Accountability	AI actors should be accountable for proper functioning of AI systems
Role-based responsibility	Accountability proportionate to role, context, and state of the art
Redress mechanisms	Enable challenge and redress for AI-driven outcomes

Penalties

Violation	Fine
Non-compliance	Voluntary — no binding enforcement mechanism