Learn AI /AI Governance and UN Terminology /AI governance
AI governance
- Starter
- AI Governance and UN Terminology
- Policy officers
- Managers & front offices
Also called: global AI governance, governance of AI.
Back to the milestone you came from
In five seconds
What it is
The rules, institutions and processes used to guide how AI is built, used and overseen so its benefits and risks are handled in the public interest.
In plain words
AI governance is the set of rules and responsibilities around AI. It covers who may use it, what checks are required, and what happens if it causes problems.
An everyday example
Imagine several neighbourhoods share one water system. Engineers build the pipes, but the system also needs rules about water quality, meters, repairs, complaints, drought plans and who pays for upgrades. Residents, city officials, maintenance teams and health inspectors all play different roles. If a pipe bursts or water is contaminated, the system needs a way to detect it, respond quickly and decide who is responsible. The water system works best not just because pipes exist, but because the surrounding rules and institutions keep it reliable and fair.
AI governance is not only about running one shared utility. AI systems are made by many actors across borders, change quickly, and are used in very different contexts such as health, education, finance and security. Unlike water quality, many AI questions involve contested values, uncertain evidence and trade-offs that cannot be solved by technical inspection alone.
Remember thisAI governance is the surrounding system of rules, responsibilities and checks for building, buying and using AI, including what happens when the system fails.
This term is contested
The broad idea is widely used, but the exact scope of AI governance is contested. Some people use it narrowly for government rule-making, while others include corporate governance, technical standards, international coordination and informal norms.
Understand it
Understand it
Why it matters
Non-specialists often meet AI governance before they meet advanced AI. It shapes what can be bought, deployed, restricted, reported and improved. If you help choose software, approve budgets, design services, draft policy or explain risks to leadership, governance determines the questions you should ask: Who is responsible? What checks exist? What evidence supports this system? How will problems be found and fixed? Without governance, AI use can become unclear, uneven and hard to challenge. With governance, decisions are more traceable, and organisations are better able to balance innovation, safety, rights and public trust.
How AI governance turns concerns into oversight
Show how evidence, values and institutions combine to produce rules, oversight and accountability for AI use.
01
Inputs
Evidence, public values, rights concerns, risks, benefits and practical needs enter the process.
Draw a rounded rectangle on the left with six small icons or labels inside: evidence, rights, risks, benefits, public needs, market reality. Use arrows pointing right.
02
Governance actions
Authorities and organisations translate those inputs into laws, policies, standards, procurement rules and review processes.
Place a larger centre box labelled governance actions. Inside, list: laws, policies, standards, audits, procurement, reporting. Connect inputs to this box with a thick arrow.
03
Implementation
Developers, buyers and deployers apply those requirements through testing, documentation, monitoring and human oversight.
Add a third box to the right labelled implementation. Inside, show four checklist items: testing, documentation, monitoring, human oversight.
04
Outcomes
The result can include safer use, clearer accountability, fewer harms, and updates when new problems appear.
Add a final box on the far right labelled outcomes and feedback. Include labels: safer use, accountability, redress, revision. Draw a feedback arrow from this box back to governance actions.
A left-to-right flow with four stages. Read from inputs, to governance actions, to implementation, to outcomes. Each stage names what enters the process, what institutions do, and what kinds of results appear in practice.
On hover or tap, each stage expands to show two examples: one public-sector example and one private-sector example.
Describe this without the diagram
This diagram shows AI governance as a four-stage flow. First, inputs enter: evidence about system performance, public values, human rights concerns, possible risks, possible benefits and practical needs. Second, governance actions convert those inputs into rules and processes such as laws, policies, standards, audits, procurement requirements and reporting channels. Third, implementation applies those requirements through testing, documentation, monitoring and human oversight by the organisations that build or use AI. Fourth, outcomes include safer use, clearer accountability, remedies when harm occurs and revision of the rules when new problems are found. A feedback arrow from outcomes back to governance actions shows that governance is continuous, not one-time.
With reduced motion: A static four-column diagram with arrows between columns and one return arrow from outcomes to governance actions. No animation; all labels and examples are visible at once.
AI governance is the practical system around AI: who sets expectations, who checks compliance, who is accountable, and how problems are addressed. It includes laws, policies, standards, oversight bodies, procurement rules, audits, reporting and ways for people to raise concerns.
What goes in? Many things: technical evidence about how a system works, reports of harm or benefit, public values, human rights concerns, security risks, market realities and the goals of governments, organisations and communities. In international settings, it also includes negotiation among Member States and input from companies, researchers and civil society.
What happens? People and institutions turn those inputs into guidance and decisions. They may write rules, agree principles, set technical standards, require testing, define responsibilities, limit certain uses, or create channels for review and redress. Governance is broader than regulation: regulation is one tool inside governance.
What comes out? Clearer expectations for developers, buyers, deployers and users of AI, plus mechanisms for oversight and correction. Good governance does not guarantee good outcomes, but it helps societies manage trade-offs, reduce harm, support useful innovation and make responsibility easier to trace when something goes wrong.
How “Running a shared water system for several neighbourhoods” maps onto the real thing
| In the story | In the AI system |
|---|---|
| Water pipes and pumps | AI systems and technical infrastructure |
| Water quality rules, billing rules and inspection routines | Policies, standards, audits and oversight requirements |
| Residents, engineers, inspectors and city officials | Users, developers, regulators, organisations and public institutions |
| Complaint and repair process after a burst pipe | Incident reporting, accountability and remedy when AI causes harm |
A worked example
An information management officer at Headquarters is finalising a concept note for an AI search assistant before clearance by an ICT governance board. The proposed pilot would search an approved collection of public reports and meeting records, returning relevant passages rather than unchecked summaries.
Starting from: Concept note: answer staff queries with links to source passages from an approved collection of public documents.
- The concept note specifies the permitted purpose, users and document collection, and records uses that remain out of scope.
- The business owner, technical owner and approving authority are named. Responsibility is assigned for testing, monitoring, incident handling and any decision to suspend the tool.
- Procurement clauses require system documentation, notice of significant model changes, access controls, usage logs and a process for investigating reported errors.
- Before deployment, staff test representative queries, misleading prompts and cases where the collection contains no answer. Responses must link to sources and state when the available evidence is insufficient.
- After launch, the owner reviews error reports and usage patterns. Material failures trigger correction, user notification or suspension of the relevant function, not simply another reminder to check the output.
Result: The cleared concept note establishes authority, procurement conditions, testing, monitoring and remediation. Together, these arrangements form the governance framework for the pilot. The search feature is only the part colleagues see.
In a United Nations workplace
By close of business, a programme officer in an evaluation office must complete the technical scoring of three bids for AI-assisted coding of interview notes and survey comments. The terms of reference require evidence on model limitations, data handling, human review, audit logs, incident reporting and correction of errors.
AI governance is already operating through the evaluation criteria. Before a preferred bidder is identified, the panel can assess safeguards, allocate responsibilities and identify conditions for the contract, including testing, documentation, monitoring, reporting, suspension and remediation.
Before you try thisBids, evaluator comments and evaluation evidence must not be entered into an AI tool unless it is approved for procurement-sensitive and personal information. An unapproved tool could expose confidential material, compromise the integrity of the evaluation and breach duties owed to interviewees or respondents.
Where you may hear it
- In General Assembly informals on draft text concerning AI
- At a technical evaluation meeting for bids containing AI features
- When an ICT governance board considers a pilot concept note
- During clearance of an impact assessment before deployment
- At a vendor demonstration when the discussion reaches accountability and incident handling
Avoid confusion
Avoid confusion
Often confused with
Regulation
A regulation is a binding rule made by a competent authority under rule-making powers granted by law for those within its scope.
AI ethics
AI ethics is the practice of checking whether AI systems are fair, safe, accountable and respectful of human rights before and during their use.
AI safety
AI safety means designing, testing and using AI systems so they are less likely to cause harm, especially in situations where people could be affected.
Side-by-side
Common misconceptions
Common misconception: AI governance just means passing an AI law.
Correction: Laws are only one part of governance. Governance also includes standards, procurement rules, oversight bodies, audits, reporting processes, professional guidance and practical ways to correct problems. An organisation can do some governance even before a new law exists.
Common misconception: AI governance is only for governments.
Correction: Governments play a major role, but organisations, companies, standards bodies, researchers and civil society also shape how AI is used and checked. A hospital, school system or UN office may have internal governance responsibilities even when national rules are still developing.
Common misconception: If a system is governed, it must be safe or fair.
Correction: Governance improves the chances of responsible use, but it does not guarantee good outcomes. Weak rules, poor enforcement, limited evidence or bad incentives can still lead to harm. Governance is a framework for managing risk and accountability, not a promise of perfection.
Quick check
Which example best shows AI governance rather than only building an AI system?
Governance is about the rules, oversight and accountability around AI use. The second option adds procurement conditions, review steps and ways to detect and respond to problems, which are governance mechanisms.
Go deeper
Go deeper
The simple version
AI governance is the set of rules and responsibilities around AI. It covers who may use it, what checks are required, and what happens if it causes problems.
The technical version
AI governance refers to the institutional, legal, procedural and operational arrangements that shape the design, development, procurement, deployment, oversight and remediation of AI systems. It can include binding and non-binding measures: legislation, regulation, standards, assurance mechanisms, impact assessment, procurement clauses, documentation duties, incident reporting, monitoring, redress and accountability allocation. In international discussion, the term may cover national governance, cross-border coordination and global processes involving states, international organisations, firms, technical communities and civil society. It is broader than regulation, broader than ethics guidance alone, and related to but distinct from AI safety, which usually focuses more narrowly on preventing specific forms of harm or failure.
Reviewed external resources
Primary sourceGlobal Digital Compact (opens in a new tab)United Nations
This United Nations page suits readers who want an official UN reference point on how AI governance is discussed within the Global Digital Compact process.
Intermediatelink checked 2026-07-31
Primary sourceGoverning AI for Humanity (opens in a new tab)UN Secretary-General's High-level Advisory Body on AI
This official report suits policy and leadership readers who want a structured UN-linked discussion of options and principles for governing AI for the public good.
Intermediatelink checked 2026-07-31
EncyclopaediaWikipedia search: AI governance (opens in a new tab)Wikipedia
This Wikipedia search results page helps a general reader see the range of articles and related topics people connect with AI governance, but it is a search page rather than one settled explanation.
search results, not one pagelink checked 2026-07-31
VideoVideo explanations: AI governance (opens in a new tab)YouTube search
This YouTube search results page suits visual learners who want to compare different video explanations of AI governance, but it is a search page with mixed sources and quality.
search results, not one pagelink checked 2026-07-31
DiscussionCommunity discussion: AI governance (opens in a new tab)Reddit search
This Reddit search results page suits readers who want to see common questions, debates and confusions about AI governance in public discussion, but it is a search page rather than an authoritative source.
search results, not one pagelink checked 2026-07-31
Where this came from
Drafted with a language model through an offline pipeline, checked against this site's schema, and every external link above fetched to see whether it answers. Treat it as a starting point and check anything that matters against the primary sources. How this site is built and what that means.
The one thing to remember
AI governance is the surrounding system of rules, responsibilities and checks for building, buying and using AI, including what happens when the system fails.