AI Regulation ReferenceDoes ISO/IEC 38507 Governance of AI require Human Oversight?
OECD • voluntary
Yes — 1 provision
Requirements at a glance
This regulation imposes 6 specific requirements for Human Oversight across 1 provision:
- Governing body responsibility — Boards and governing bodies must evaluate, direct, and monitor the organisation's use of AI
- Effective use — Ensure AI is used effectively to fulfil organisational objectives
- Efficient use — Ensure AI use delivers value proportionate to resources and risks
- Acceptable use — Ensure AI use complies with applicable laws, regulations, and ethical expectations
- AI governance framework — Establish governance structures for oversight of AI across the organisation
- Accountability assignment — Assign clear accountability for AI-related decisions and outcomes at executive level
Board-Level AI Governance #
ISO/IEC 38507 is the only international standard specifically addressed to governing bodies (boards, executives) rather than technical teams — directing boards to evaluate, direct, and monitor AI use. As regulators increasingly hold organisations accountable at the board level for AI governance, this standard defines what board-level AI oversight looks like.
Requirements
| Requirement | Details |
|---|---|
| Governing body responsibility | Boards and governing bodies must evaluate, direct, and monitor the organisation's use of AI |
| Effective use | Ensure AI is used effectively to fulfil organisational objectives |
| Efficient use | Ensure AI use delivers value proportionate to resources and risks |
| Acceptable use | Ensure AI use complies with applicable laws, regulations, and ethical expectations |
| AI governance framework | Establish governance structures for oversight of AI across the organisation |
| Accountability assignment | Assign clear accountability for AI-related decisions and outcomes at executive level |
Penalties
| Violation | Fine |
|---|---|
| Non-compliance | Voluntary — no binding enforcement mechanism |