Global AI Governance Frameworks
Most global AI governance is voluntary, with one binding exception. The Council of Europe Framework Convention on AI, opened for signature on 5 September 2024, is the first legally binding international AI treaty. The EU ratified it in May 2026, but it was not yet in force as of mid-2026.
The voluntary frameworks still shape national laws. The OECD AI Principles are the most widely adopted baseline. The G7 Hiroshima AI Code of Conduct sets eleven actions for advanced AI developers, and the UNESCO Recommendation on the Ethics of AI covers human rights and sustainability. The Bletchley Declaration of November 2023 launched the international AI safety summit series.
The UN now runs two AI processes created in August 2025. The Global Dialogue on AI Governance first met in Geneva in July 2026, and an independent scientific panel published its first report the same month. Sector bodies such as the Financial Stability Board and FATF publish non-binding guidance on AI in finance and anti-money laundering.
Key themes
- 1.Council of Europe AI Convention, the first binding AI treaty
- 2.OECD AI Principles as the global voluntary baseline
- 3.G7 Hiroshima Code of Conduct and the AI safety summit series
- 4.UN Global Dialogue and Scientific Panel on AI
Regulatory frameworks and guidance(12)
Bletchley Declaration on AI Safety
At the inaugural AI Safety Summit, 28 governments signed this political declaration. They recognized risks from frontier (most advanced) AI and committed to international cooperation on safety, evaluation, and information sharing.
Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law
This Council of Europe treaty is the first legally binding international instrument on AI. It covers AI used by public authorities and private actors acting for them; Parties decide how to address other private actors. It opened for signature on 5 September 2024 and was not yet in force as of mid-2026.
FATF AI Anti-Money Laundering Guidance
A 2021 Financial Action Task Force (FATF) report on new technologies, including AI, for fighting money laundering and terrorist financing. It sets out benefits, challenges, and suggested actions for governments, and is non-binding.
Sound Practices for Responsible Adoption of Artificial Intelligence (Consultation Report)
The Financial Stability Board proposes 12 practices for responsible AI adoption throughout its lifecycle. They cover banks, insurers, and other regulated financial entities developing or deploying AI. Institutions should map them to governance, model risk, third-party oversight, and lifecycle controls.
Five Eyes Guidance on the Careful Adoption of Agentic AI Services
This joint Five Eyes advisory addresses enterprises and public bodies deploying autonomous AI agents. It covers agents taking independent actions, accessing systems, or interacting with other agents. The guidance calls for low-risk tasks, minimum necessary access, and integration into existing security governance.
G7 Hiroshima AI Code of Conduct
The G7 Hiroshima AI Process issued this voluntary international code of conduct. It sets eleven principles and corresponding actions for advanced AI developers and operators, particularly frontier (most capable) and general-purpose models. The code uses voluntary commitments to guide responsible development.
ITU Focus Group on Trust and Identity for Humans and Agentic AI
The International Telecommunication Union (ITU) launched a Focus Group on trusted digital identity and accountable behavior throughout the lifecycles of AI agents. It addresses agent identification, credentials, authorization, and agent-to-agent interactions. The work concerns organizations developing or deploying agents with delegated authority or external system access.
OWASP Top 10 for Large Language Model Applications
OWASP's Top 10 for LLM Applications lists the most critical security risks in applications built on large language models. The current 2026 edition, published in August 2026, puts prompt injection, sensitive information disclosure, and excessive agency at the top. Development and security teams use it to prioritize safeguards.
The Role of Investors in AI Governance
This April 2026 research paper from Oxford's AI Governance Initiative reports how institutional investors approach AI governance. It draws on six roundtables with 62 investors and 60 interviews, mostly about public equities. It describes current practice and the support investors want, rather than setting requirements.
Singapore Consensus on Global AI Safety Research Priorities
The Singapore Consensus is a research agenda for AI safety, written by an expert committee after a scientific exchange Singapore hosted in April 2025. It groups research priorities into development, assessment, and control. It is not a government declaration, has no signatories, and was updated by a 2026 edition.
Global Dialogue on AI Governance (UN General Assembly Resolution A/RES/79/325)
UN Resolution A/RES/79/325 established the Global Dialogue on AI Governance, open to states and other stakeholders. The first Dialogue met in Geneva on 6 and 7 July 2026, informed by 1,532 written submissions. The second is set for New York on 3 and 4 May 2027.
Verifiable Semiconductor Manufacturing
An April 2026 research paper from Oxford's AI Governance Initiative on verifying the integrity of AI chips from design through manufacturing. It argues for defense in depth, tiered by chip criticality, with neutral third-party verifiers. It sets no requirements for anyone.