OECD AI Principles
// Intergovernmental Principles - The Common Ancestor
Voluntary on paper, foundational in practice. The OECD AI Principles are where governments first agreed what trustworthy AI means - and their definitions now sit inside binding statutes like the EU AI Act. Align here and you are pre-aligned with much of what regulators write.
Key Highlights
Why a non-binding instrument carries legal weight downstream.
- 01
The OECD AI Principles, adopted May 2019 and updated May 2024, are the first intergovernmental AI standard - the common ancestor most national frameworks openly cite.
- 02
Five values-based principles: inclusive growth and well-being, human-centred values, transparency and explainability, robustness and safety, and accountability.
- 03
Five recommendations to governments: investing in AI research, fostering an inclusive AI ecosystem, shaping an enabling policy environment, building human capacity, and international cooperation.
- 04
The 2024 update extended coverage to general-purpose AI models and added expectations around safety incident tracking - anticipating where binding laws then went.
- 05
The OECD definition of an "AI system" was adopted nearly verbatim by the EU AI Act, giving these principles unusual legal gravity for a voluntary instrument.
- 06
The OECD.AI Policy Observatory tracks over 1,000 policy initiatives across 70+ jurisdictions, making it the single best free radar for regulatory movement.
Action Items for Governance and Compliance
Four moves that convert soft-law alignment into hard compliance readiness.
Anchor corporate AI values to the five principles
A01Have the board endorse a responsible-AI charter that references the principles explicitly.
Maintain a principle-to-control mapping reviewed annually so values demonstrably shape controls.
Adopt shared definitions in internal registers
A02Use the OECD taxonomy in your system inventory so reports interoperate with regulators.
Reuse the OECD AI-system definition when classifying estates - it matches what EU law will ask.
Address general-purpose AI expectations
A03Assign ownership for evaluating systemic risk when adopting foundation models.
Log safety incidents in a format consistent with updated OECD guidance to stay ahead of reporting mandates.
Monitor policy through the OECD Observatory
A04Give one team standing responsibility for quarterly regulatory radar reviews.
Feed findings into roadmap decisions so obligations are anticipated rather than absorbed.
Engineering Translation
High-level capabilities that make principle alignment durable.
>_Shared vocabulary baked into registries so internal reports interoperate with regulator taxonomies.
>_Principle tags attached to controls making cross-framework mapping automatic instead of manual.
>_Incident records structured once, consumable by every future reporting regime.
Delivered as capabilities, not paperwork
Each obligation above is translated into product-level engineering capabilities. The underlying stack and internal tooling are intentionally abstracted here; they are covered in technical briefings.
Educational summary only. This is not legal advice. Confirm obligations with qualified counsel for your jurisdiction.
Align With the Source Most Regulators Quote
Book a technical briefing to map your governance controls against the OECD principles and their statutory descendants.