Hael
Sign inBook a demo
All frameworks
INTERNATIONALOECD/LEGAL/0449REFERENCE

OECD AI Principles — the foundation every national AI strategy references.

How Hael produces the AI Principles alignment documentation national regulators increasingly reference — generated against the operational controls each principle implies.

2019
Adopted
Updated
May 2024
47
Adherent jurisdictions
Foundation
For national strategies

What the OECD AI Principles establish

The OECD AI Principles, adopted in 2019 and updated in May 2024 to address generative AI, are the most widely subscribed international reference standard for trustworthy AI. Forty-seven jurisdictions adhere to the Principles, and they form the foundation for national AI strategies including the US AI Bill of Rights, EU AI Act, UK AI policy framework, Canada AIDA, Japan AI Strategy and the G7 Hiroshima AI Process.

The Principles establish five values-based principles: inclusive growth, sustainable development and well-being; human-centred values and fairness; transparency and explainability; robustness, security and safety; and accountability. They also establish five recommendations to policymakers covering investment in AI R&D, fostering a digital ecosystem, shaping an enabling policy environment, building human capacity, and international cooperation. The Principles are voluntary but have substantial soft-law influence; alignment documentation is increasingly referenced in regulatory engagement and procurement.

How Hael runs it

Hael maps each Principle to concrete operational controls running on the platform. Inclusive growth maps to the benefits assessment surface and stakeholder engagement records. Human-centred values maps to FRIA, bias evaluation and oversight assignment. Transparency maps to documentation generation, model cards and explainability surfaces. Robustness maps to evaluation pipeline, adversarial testing and incident response. Accountability maps to ownership assignment, audit chain and governance decision records.

An OECD alignment artefact is generated per organisation and per AI system — describing how each Principle is operationalised, referencing the substantive evidence and sealed with cryptographic provenance. Regulators and procurement teams referencing the OECD Principles receive a concrete artefact, not a generic attestation.

PrincipleElementCoverageHow Hael runs it
1.1Inclusive growth, sustainable developmentFullBenefits assessment and stakeholder engagement records
1.2Human-centred values and fairnessFullFRIA, bias evaluation, oversight assignment
1.3Transparency and explainabilityFullDocumentation generation, model cards, explainability surfaces
1.4Robustness, security and safetyFullEvaluation pipeline, adversarial testing, incident response
1.5AccountabilityFullOwnership assignment, audit chain, governance decision records
1.5GenAI overlay — content provenanceFullWatermarking and synthetic-content notice scaffolding
1.5GenAI overlay — training data transparencyFullTraining data lineage and summary generation
ALIGN-1Per-system alignment artefactFullGenerated artefact with sealed provenance per agent
ALIGN-2Per-organisation alignment statementFullAggregated alignment statement for regulator engagement
ALIGN-3Cross-framework referenceFullMaps onto NIST AI RMF, ISO 42001, EU AI Act controls

Questions

Are the OECD AI Principles enforceable?

The Principles are voluntary international soft law. They do not carry direct enforcement, but adherent jurisdictions have integrated them into binding national frameworks. Alignment documentation is increasingly referenced in regulatory engagement, procurement and international AI governance forums.

How do the Principles relate to the EU AI Act and NIST AI RMF?

Both the EU AI Act and NIST AI RMF cite the OECD Principles as foundational. The Act's risk-based architecture, transparency obligations and fundamental rights focus all reflect Principle 1.2 and 1.3. NIST AI RMF's trustworthy AI characteristics are explicitly aligned with the Principles.

What did the May 2024 update add?

The 2024 update extended the Principles to generative AI specifically — addressing content provenance, training data transparency, misuse risks and the safety considerations distinctive to general-purpose models. The update was coordinated with the G7 Hiroshima AI Process.

See Hael produce your OECD alignment artefact.

Per-Principle operational mapping, per-agent alignment artefact, per-organisation alignment statement — sealed with hash-chained provenance.

Book a demoSee the platform