China Anchors a New AI Governance Bloc in Shanghai

China now has its own intergovernmental institution for setting the rules of artificial intelligence. On July 16, 2026, representatives of 29 governments signed an agreement in Shanghai establishing the World Artificial Intelligence Cooperation Organization, or WAICO, an independent intergovernmental body that will be headquartered in the city — and no G7 economy is among its founding members.

Chinese Foreign Minister Wang Yi signed on behalf of the Chinese government, with United Nations Secretary-General António Guterres among those present at the ceremony, held on the eve of the annual World Artificial Intelligence Conference. The signing turns a proposal that Premier Li Qiang floated at the 2025 edition of that conference into a constituted organization with a founding charter, a permanent seat, and a defined membership. It is the clearest sign yet that Beijing intends to build AI governance institutions, not just export AI hardware and standards.

What the agreement actually establishes

According to the official Chinese account, the agreement commits WAICO to the purposes of the UN Charter, a principle Beijing frames as extensive consultation, joint contribution, and shared benefit, and a “people-centered” approach to AI. Its stated purpose is to promote international cooperation and global governance of the technology, and to steer it toward outcomes the account calls beneficial, safe, and fair.

Those are aims, not obligations. What the signing does not yet make public is the machinery that would let WAICO act on any of them. Chinese officials say the founding text covers the organization’s purpose, principles, functions, structure, decision-making procedures, and membership — but the operative details that decide whether an international body has any teeth have not been released: how members vote, who pays for it, what it can require of a member, and how a state joins or leaves. Signing the agreement establishes the institution on paper; on the public record, it does not yet make it operational.

That gap is the substance of the story, not a footnote to it. In an analysis written for the Carnegie Endowment for International Peace before the signing, governance researcher Arindrajit Basu argued that WAICO’s functional details remained unclear, and that Beijing would most likely use it to build a preliminary consensus among members and then carry that consensus into United Nations forums, where China increasingly sees room to shape the agenda. The organization traces back to a thirteen-point global AI governance action plan China released alongside the original 2025 proposal, a document that leaned heavily on capacity-building and infrastructure for developing countries.

Who signed, and who stayed out

The founding roster is a geopolitical signal in itself. Alongside China, the signatories include Russia, Indonesia, Brazil, Pakistan, Malaysia, South Africa, and Kazakhstan — a membership drawn overwhelmingly from the Global South, together with Russia and its partners. Absent are the United States, the European Union member states, the United Kingdom, Japan, and the other economies that anchor the Western-led AI governance forums and are drafting their own sovereign-AI agendas.

That absence defines what WAICO is positioned against. The established governance venues — national safety institutes, the G7’s process, the OECD’s principles — largely operate as clubs organized around shared values and the management of frontier-AI risk. WAICO inverts that model: membership turns on being a sovereign state rather than on passing a values test, and its agenda foregrounds development and access over rules and risk. It also arrives inside the US–China contest over AI, as Washington tightens export controls on advanced chips and, by Basu’s account, pulls back from the multilateral norms-setting venues it once led.

What Beijing is building toward

At the conference the next day, Chinese President Xi Jinping used his keynote to cast Beijing as the convener of an open, cooperative order. “AI development should not be a solo performance by a single country, but a symphony of international cooperation,” he said, warning against any country “overstretching the national security concept” in AI — language widely read as aimed at US chip restrictions.

For the companies and governments watching, the immediate effect is diplomatic rather than regulatory. WAICO issues no binding rules today, and its influence will depend on what its members do next. The signals worth tracking are concrete:

• whether the founding agreement is ratified into force and a secretariat and budget actually materialize
• whether the first funded programs do anything to close the “AI divide” the organization invokes
• how it positions itself relative to the UN’s own AI track, which China publicly calls the main channel for global rules

A charter and a headquarters give Beijing a standing platform in a debate that has so far run largely through Western institutions and through a technology field where oversight has trailed capability. Whether that platform ends up writing rules or mainly coordinating a caucus of the Global South is the question its founding documents leave open.