Putting the OECD AI Principles into practice
OECD AI Principles (revised 2024)
- Status
- Voluntary
- Binding
- No
- Object analysed
- AI system
- Requirements
- 5
- Next milestone
- —
In short
Common baseline adopted by 47 countries, source of the AI Act and NIST definitions. 5 values-based principles and 5 recommendations to governments.
Steps to compliance
- Inventory your AI systemsThis framework does not classify systems: its principles apply to the whole organisation.
- Determine your roleDuties vary by role: AI actor.
- Apply the 5 requirementsThey focus on: Environmental impact, Fairness & non-discrimination, Personal data and Transparency & notice.
- Prove it with checks7 checks to document, 7 of which also serve NIST AI RMF, UK principles and AI Promotion Act.
- Keep compliance up to dateEvery known milestone has passed: the obligations apply.
Scope and penalties
- Kind
- Principles
- Scope
- AI actors.
- Territorial reach
- Adherent countries.
- Penalties
- None.
- Jurisdiction
- International
Timeline
Qualifying a system
Classification axes and possible verdicts
No classification: a principles framework applied to the whole organisation.
Requirements
5 requirements
| Code | Article | Requirement | Applies to | Checks |
|---|---|---|---|---|
| OECD-1.1 | 1.1 | Inclusive growth, sustainable development, well-being | ||
| OECD-1.2 | 1.2 | Human rights, democratic values, fairness, privacy | ||
| OECD-1.3 | 1.3 | Transparency and explainability | ||
| OECD-1.4 | 1.4 | Robustness, security and safety | ||
| OECD-1.5 | 1.5 | Accountability |
Checks to document
Evidence collected for a check counts for every regulation that uses it.
| Code | Check | Scope | Themes | Also used by |
|---|---|---|---|---|
| CHK-ENV | Environmental impact and sustainability of model training are assessed | Model | ||
| CHK-BIAS | Fairness and bias are evaluated and results documented | System | ||
| VER-019-D-01 | Need for a DPIA assessed | System | ||
| CHK-TRANSPARENCY | Transparency and accountability risks are examined and documented | System | ||
| CHK-SAFETY | Safety risks are evaluated and safe-failure demonstrated | System | ||
| CHK-SECURITY | Security and resilience are evaluated and documented | System | ||
| CHK-ROLES-CLARIFIED | Roles, responsibilities and delegated authorities are documented and clear to relevant stakeholders | Organisation |
Themes covered
Frequently asked questions
Who is in scope of OECD AI?
AI actors. Adherent countries.
What penalties does OECD AI carry?
None.
When do the OECD AI obligations apply?
May 3, 2024: Revision (AI system definition); Feb 7, 2025: Hiroshima Process reporting framework.
Is OECD AI binding?
No. Kind: principles. Status: voluntary.
How does OECD AI relate to other regulations?
The same checks serve several texts. Shared checks: NIST AI RMF (7), UK principles (3) and AI Promotion Act (2).
Related regulations
Official sources
Data checked on Sep 25, 2026. General information, not legal advice. Check the official texts and get advice for your situation.
Run these requirements across all your AI systems
TrustFlow inventories your systems, qualifies them under each regulation and collects evidence once for every referential.