Privacy Map

Organisation for Economic Co-operation and Development (OECD). AI / automated decisions. Recommendation (OECD Council)

OECD AI PrinciplesRecommendation of the Council on Artificial Intelligence (OECD AI Principles)

In force

§1Summary

Synopsis

Five values-based principles for trustworthy AI: inclusive growth and wellbeing; human rights and democratic values including privacy and fairness; transparency and explainability; robustness, security and safety; and accountability. These are paired with five recommendations to governments. The definition of an 'AI system' it contains was adopted in the EU AI Act.

Key themes
  • trustworthy AI
  • human rights
  • privacy
  • transparency
  • explainability
  • robustness
  • accountability
Unique features

It was the first intergovernmental AI standard. Its AI-system definition was taken up in the EU AI Act and the CoE AI Convention.

SaaS relevance

Soft-law benchmark for AI governance in SaaS products, and the source of definitions used in binding laws.

§2Obligations

Breach notification
None
DPO
No
Cross-border transfers
None
Maximum penalty
None

§3Dates

Key dates
  • Adopted
  • In force
  • OECD/LEGAL/0449; adopted 2019-05-22; updated 2024-05-03 (generative AI, the definition of an AI system, safety, information integrity); G20 AI Principles (2019) draw on it

§4Text and authority

Text
Authority
OECD Working Party on AI Governance / OECD.AI

§5Where it applies

Members of the body

Membership-based — check ratification status. Membership of Organisation for Economic Co-operation and Development (OECD) does not mean a state has signed or ratified this instrument.

§6Sources and verification

Sources consulted
Verification

High confidence.