Products, laws and numbers here change fast. Checked October 2026.
In plain words
When people talk about controlling AI, they often mean laws. Laws are one part. AI governance is the bigger picture: everything that shapes how AI is built, tested, sold and used, and who checks it.
It works at three levels. Inside a company: policies, review boards, risk checks and a named person in charge. In a country: laws, regulators, courts and official guidelines. Between countries: shared principles, treaties, summits and scientific panels.
Most of it is not binding law. Standards and frameworks are often voluntary, and international agreements set direction more than rules. That makes governance flexible, but uneven. As of October 2026, the European Union has a single, wide-ranging AI law, while India and the United States lean more on guidelines and existing laws.
Everyday example
How a cricket match is governed
A cricket match has the Laws of Cricket, but that is not all that keeps it fair. Umpires apply the rules on the field. A third umpire checks replays. Teams have their own codes of conduct. A national board picks players and punishes misconduct. An international council sets tournament rules. AI governance is the same mix: written laws, plus the people, checks and organisations that apply them at every level.
Where the example breaks: Cricket has one world body that everyone accepts. AI has no single global authority, and countries disagree on how strict the rules should be, so governance is a patchwork.
How it actually works
Here are the main layers, with dates, as of October 2026.
Inside organisations
Company principles, risk reviews, testing and documentation, often called responsible AI. A widely used guide is the US NIST AI Risk Management Framework (January 2023), which is voluntary and built around four functions: Govern, Map, Measure and Manage.
National rules
Laws and guidelines, covered on our AI regulation page. The EU AI Act sorts uses by risk. India's AI Governance Guidelines (November 2025) favour existing laws plus new bodies: an AI Governance Group, a technology and policy expert committee, and an AI Safety Institute.
Shared principles
The OECD AI Principles (May 2019, updated May 2024) have 47 adherents, and the G20's AI principles are drawn from them. UNESCO's Recommendation on the Ethics of AI was adopted by 193 countries in November 2021. These guide governments but do not bind them.
A treaty
The Council of Europe's Framework Convention on AI, which it calls the first legally binding international treaty on AI, opened for signature in September 2024. Signatories include the EU, the UK and the US. It enters into force after five ratifications, including three by Council of Europe member states. The EU ratified it in May 2026.
Summits and the UN
Leaders' summits began at Bletchley Park in 2023, followed by Seoul (2024), Paris (2025) and the AI Impact Summit in New Delhi (February 2026). In August 2025 the UN General Assembly created an Independent International Scientific Panel on AI and a Global Dialogue on AI Governance, which first met in Geneva in July 2026.
- Governance vs regulation
- Regulation is the legally binding part. Governance also includes voluntary standards, company policies and international cooperation
- UN science panel
- 40 members, created by resolution A/RES/79/325 on 26 August 2025. It assesses AI's risks and benefits but does not set rules
- India
- India AI Governance Guidelines, MeitY, 5 November 2025, built on seven principles it calls sutras
- Checked
- 31 October 2026. Bodies and agreements change often: check official sources
Shared scientific assessments, such as the International AI Safety Report, are part of governance too: they give governments a common picture of the evidence.
Questions people ask
What is AI governance in simple words?
All the ways we steer how AI is built and used: laws, government guidelines, technical standards, company policies, and agreements between countries, plus the people and bodies who check that they are followed.
What is the difference between AI governance and AI regulation?
AI regulation means binding laws and rules from governments. AI governance is wider: it includes regulation, but also voluntary standards, company policies, oversight boards and international cooperation. Regulation is one tool inside governance.
Is there a global body that governs AI?
Not one with power to set binding rules, as of October 2026. The UN has a scientific panel and a Global Dialogue on AI Governance, the OECD and UNESCO set shared principles, and the Council of Europe has a treaty. Binding rules still come mostly from individual countries and the EU.
What does an AI governance framework include?
Usually: principles, clear roles and accountability, a way to list and rate AI systems by risk, testing and documentation before launch, human oversight, monitoring after launch, and a process for incidents and complaints. NIST's framework groups this as Govern, Map, Measure and Manage.
How does India govern AI?
Through guidelines and existing laws rather than one AI act, as of October 2026. The India AI Governance Guidelines of November 2025 set seven principles and propose new bodies. The IT Act, the DPDP Act and 2026 IT Rules on labelling AI content apply alongside.