What AI governance actually is
AI governance is an organisation's approach to using laws, policies, frameworks, practices and processes at international, national and organisational levels. The exam loves the distinction → Principles are the values, Frameworks operationalise them.
AI governance is an organisation's approach to using laws, policies, frameworks, practices and processes at international, national and organisational levels. It helps stakeholders implement, manage, oversee and regulate the development, deployment and use of AI, manages associated risks, and keeps AI aligned with stakeholder objectives, responsible, ethical and compliant. Its guardrails address bias, privacy impacts and misuse while increasing innovation and trust.
| Principles | Frameworks | |
|---|---|---|
| What | A set of values → guidelines enabling consistency, standardisation and responsible use. Similar around the world. | A means to operationalise those values → context-sensitive, fit for specific purposes, never one-size-fits-all. One framework can be aligned to in multiple ways. |
| Examples | OECD AI Principles · FIPs · UNESCO Recommendation on the Ethics of AI | ISO 42001 (AI management system) · ISO/IEC 22989 (terminology) · NIST AI RMF (US) · IEEE 7000-2021 (ethics in system design) · HUDERIA (Council of Europe) |
If the question contrasts the two → principles = the values, frameworks = the operationalisation. UNESCO and OECD sit on the principles side; NIST, ISO and HUDERIA on the framework side.