Learn AI /AI Governance and UN Terminology /Multi-stakeholder governance

Multi-stakeholder governance

Also called: multistakeholder.

Detail

In five seconds

What it is

A way of shaping rules or shared practices with several kinds of actors, such as governments, companies, experts, civil society and affected communities.

In plain words

It means making AI-related rules or guidance with input from several kinds of people and organisations, not only governments.

An everyday example

A popular online game is changing its community rules. The company convenes players, volunteer moderators, parents, accessibility advocates and security specialists. Players report patterns of harassment, moderators identify rules that are difficult to enforce, advocates flag barriers, and specialists explain technical limits. A small drafting group publishes proposed changes and a response table, then revises the text after comments. The company still has formal authority to adopt the rules. The process is multi-stakeholder because different groups have structured ways to shape the result, not merely because they were invited to a call.

Where the story stops working

A library usually has one clear authority that can make the final rule. In real AI governance, authority is often split across countries, regulators, companies and international bodies. Some outputs are voluntary, not binding. Power is also less balanced than the story suggests, because some participants can afford far more staff, travel, data and lobbying access than others.

Remember thisCount influence, not invitations: multi-stakeholder governance depends on how different actors can shape the agenda, evidence, drafts and revisions, and on who ultimately decides.

This term is contested

The broad idea is widely used, but the term is contested in practice because people disagree about which stakeholders should be included, how much influence non-state actors should have, and whether some models give disproportionate power to large firms or well-resourced groups.

Understand it

Why it matters

Decisions about AI can affect rights, safety, markets, public services and international cooperation. No single group sees the whole picture. A multi-stakeholder process can bring in technical knowledge, lived experience and implementation realities that a state-only or company-only process might miss. That can improve legitimacy, spot harms earlier and make rules easier to apply across borders. But participation alone is not enough. If some voices are excluded, under-resourced or heard too late, the process may look inclusive without being balanced. For non-specialists, the key question is not just who was in the room, but who had influence, at what stage, and over what kind of outcome.

How multi-stakeholder governance turns many inputs into a shared output

After viewing this, the reader should be able to explain that the approach combines different actors, evidence and negotiation processes to produce guidance, standards or draft rules.

01
Stakeholders enter
Different groups join the process: states, companies, technical experts, civil society, researchers and affected communities.
Draw six labelled icons or boxes in one column. Use equal box size to avoid implying equal power, but add a note that influence may still differ in practice.
02
Evidence and interests are shared
Participants bring legal concerns, technical knowledge, operational experience, rights impacts and implementation constraints.
Merge arrows from all stakeholder boxes into a shared tray labelled 'evidence, concerns, proposals'.
03
Text and options are negotiated
Through consultations, workshops or drafting groups, participants compare proposals, revise wording and identify trade-offs.
Place a central process box labelled 'consultation, drafting, negotiation' with looping arrows inside to show revision.
04
Outputs are produced
The process may produce recommendations, standards, codes of practice, frameworks or language that informs state-led decisions.
Draw an output column with five small boxes: 'report', 'standard', 'guidance', 'code of practice', 'draft language for policy'.

A left-to-right flow diagram. The first column shows stakeholder groups. Arrows lead to a middle column for consultation and negotiation steps. Arrows then lead to an output column showing different kinds of governance result. Read from left to right to see who joins, what they do together and what comes out.

Hovering or tapping a stakeholder highlights the kinds of input that actor commonly contributes and shows that all inputs pass through the same negotiation stage before any output appears.

Describe this without the diagram

This diagram explains multi-stakeholder governance as a flow with four stages. First, several kinds of participants enter the process: states, companies, technical experts, civil society, researchers and affected communities. Second, they contribute different inputs such as evidence, concerns, proposals and implementation limits. Third, these inputs move into a shared negotiation stage, where consultation, drafting and revision happen. Fourth, the process produces outputs such as reports, standards, guidance, codes of practice or draft language that may later inform formal policy decisions. The main message is that the term describes a governance process involving multiple kinds of actors, not a single legal instrument.

With reduced motion: A static left-to-right chart with four labelled columns: stakeholders, shared inputs, negotiation, and outputs. No animation. Arrows remain visible at all times to show the process path.

In this approach, more than one kind of actor takes part in shaping how an issue is handled. The usual groups are governments, businesses, technical experts, civil society organisations, academics and sometimes people directly affected by the issue. In AI, the issue might be safety testing, standards, access, human rights impacts or cross-border interoperability.

What goes in are different interests, kinds of knowledge and forms of authority. A company may know how a system is built. A regulator may know legal duties. A civil society group may bring evidence about harms. Researchers may explain technical limits. The process then uses meetings, consultations, working groups, draft texts or codes of practice to compare views and negotiate language.

What happens next depends on the forum. Sometimes the output is advice, a voluntary standard, a report or a framework. Sometimes it helps states prepare a resolution, regulation or other normative instrument. So this term does not mean one specific legal tool. It describes who participates in governance and how participation is organised.

It is often confused with multilateralism, but they are not the same. Multilateralism mainly means cooperation among states. Multi-stakeholder governance includes states but also non-state participants. It also does not guarantee equal power. Some actors may still have more money, access, technical capacity or agenda-setting influence than others.

How “Rewriting the rules of an online game” maps onto the real thing

In the story	In the AI system
Library director and invited groups	Governments and non-state stakeholders taking part in governance
Different problems raised by each group	Different evidence, interests and expertise entering the process
Several meetings and a draft policy	Consultations, working groups and negotiated text
Booking rules, safeguards and review date	Outputs such as frameworks, standards, guidance or draft regulatory language

A worked example

A standards focal point in a Bogotá technical secretariat must circulate draft guidance on synthetic voice tools in public-facing services before the next plenary.

Starting from: The process brings together governments, technology providers, technical experts, disability advocates, child-protection organisations, researchers and representatives of people who use accessible public services.

The secretariat invites written evidence on authentication, accessibility, fraud, freedom of expression and interoperability.
The focal point prepares a synthesis that identifies areas of convergence, competing proposals and gaps in the evidence.
A mixed drafting group converts the submissions into draft provisions. Participants then comment on the text and explain objections or proposed amendments.
The secretariat circulates a revised draft with a response table showing how substantive comments were addressed. The designated technical committee considers the text at plenary.

Result: The committee issues non-binding guidance. Several kinds of actor shaped the agenda and successive drafts, while formal authority for the final output remained with the committee.

In a United Nations workplace

A political affairs officer in a specialised agency technical division is finishing a handover note for an incoming colleague before close of business. The note maps who may submit evidence, join drafting sessions and comment on revisions to proposed guidance on AI assurance. It also records that formal approval remains with the agency’s governing body.

The governance question is not simply who appeared on the participant list. The handover note must show which actors could set the agenda, propose text, challenge evidence or respond to revisions, and which body retained authority over the outcome.

Before you try thisDo not present an open consultation as joint decision-making if outside participants had no drafting or approval role. Unpublished positions, internal instructions and sensitive correspondence must not be entered into an external AI tool unless that tool has been approved for the material.

Where you may hear it

A consultation briefing in the margins of an Economic and Social Council session
A first reading of draft guidance in an International Telecommunication Union technical meeting
An inter-agency working group reviewing comments from civil society and industry
A side event on implementing ethics guidance for artificial intelligence

Avoid confusion

Often confused with

Common misconceptions

Common misconception: It means everyone gets an equal vote.
Correction: Not necessarily. Many multi-stakeholder processes do not use formal voting at all, and even when participation is broad, influence can still be uneven. Governments may retain legal authority, while firms or well-funded organisations may have more staff, data or access than smaller groups.
Common misconception: It is just another word for multilateralism.
Correction: Multilateralism usually refers to cooperation among states. Multi-stakeholder governance includes states but also non-state actors such as companies, experts and civil society. The two can overlap, but they answer different questions about who participates.
Common misconception: If a process is multi-stakeholder, the result is automatically legitimate and fair.
Correction: Inclusion can improve legitimacy, but it does not guarantee it. A process may still exclude affected communities, set the agenda narrowly, rely on opaque drafting, or privilege actors with more money and technical capacity.

Quick check

Which statement best captures multi-stakeholder governance in AI?

Go deeper

The simple version

It means making AI-related rules or guidance with input from several kinds of people and organisations, not only governments.

The technical version

Multi-stakeholder governance is an institutional arrangement for developing or influencing governance outputs through structured participation by heterogeneous actors, typically including states, private sector entities, technical communities, civil society, academia and sometimes directly affected groups. The mechanism usually combines consultation, agenda-setting, evidence submission, drafting and iterative revision. Its outputs may be binding or non-binding depending on the forum, and can include standards, codes of practice, policy frameworks, advisory reports or text that later feeds into formal intergovernmental or domestic regulatory processes. The term therefore describes the composition and procedure of governance, not the legal status of the outcome. It is distinct from purely intergovernmental processes, although many real cases are hybrid arrangements where states retain final authority while non-state actors shape technical substance, implementation design or legitimacy.

Reviewed external resources

Primary sourceGlobal Digital Compact (opens in a new tab)United Nations
This UN page shows how the concept appears in an official digital cooperation context and suits readers looking for current UN framing.
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 report gives a substantive UN-linked discussion of AI governance arrangements and suits policy readers who want broader context and recommendations.
Intermediatelink checked 2026-07-31
EncyclopaediaMultistakeholder governance (opens in a new tab)Wikipedia
This overview gives a general starting point on the term and suits readers who want a quick orientation before reading formal documents.
Intermediatelink checked 2026-07-31
VideoVideo explanations: Multi-stakeholder governance (opens in a new tab)YouTube search
This is a search results page for videos, so it helps readers find visual explainers from different creators and suits people who learn better by watching.
search results, not one pagelink checked 2026-07-31
DiscussionCommunity discussion: Multi-stakeholder governance (opens in a new tab)Reddit search
This is a search results page for community discussion, so it helps readers see common questions and debates and suits readers comparing public interpretations.
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

Count influence, not invitations: multi-stakeholder governance depends on how different actors can shape the agenda, evidence, drafts and revisions, and on who ultimately decides.