China Is Not Just Building AI. It Is Trying to Write the Rules Around It.
When Xi Jinping spoke at the World AI Conference in Shanghai in July 2025, his description of AI development as a “symphony” of international cooperation sounded familiar enough. Four months later, however, the language had acquired an institutional form. At the Fourth Plenum, China put forward a proposal for a World AI Cooperation Organization, or WAICO, to be headquartered in Shanghai.
That sequence matters.
China was not simply adding another initiative to the growing list of international discussions about artificial intelligence. It was proposing an institution of its own at a moment when the rules governing AI remain unsettled.
And China has been here before.
Over the past two decades, Beijing has helped build institutions ranging from the Asian Infrastructure Investment Bank and the New Development Bank to the Shanghai Cooperation Organisation and the expanded BRICS framework. In 2023, it also launched its Global AI Governance Initiative.
Taken separately, these belong to very different parts of international affairs. Taken together, they suggest a broader strategy: when existing institutions do not give China enough influence over an emerging domain, Beijing increasingly creates additional institutions through which that influence can accumulate.
AI may be the most consequential test of that strategy yet.
The shift from joining institutions to building them
China already has substantial institutional power.
It has a permanent seat on the UN Security Council. It is one of the world’s largest economies. It has enormous weight in global trade and finance.
But many of the institutions that govern those systems were designed before China acquired its present economic and technological weight.
Changing them from within is possible, but slow.
Building additional institutions offers another route.
The Asian Infrastructure Investment Bank is an early example. Established in 2016, it gave China a founding role in a multilateral development institution with substantial capital and a membership that extended well beyond Beijing’s traditional partners. Several US allies joined despite Washington’s objections.
That episode mattered for a reason that went beyond infrastructure finance. It demonstrated that countries could participate in a China-backed institution without automatically abandoning their existing relationships with the United States and its allies.
The New Development Bank took a different route, placing an alternative development-finance institution within the BRICS framework.
BRICS itself subsequently expanded beyond its original five members. The Shanghai Cooperation Organisation developed into a broader platform for political and security cooperation.
None of these institutions displaced the World Bank, the IMF, the G7 or the established security architecture.
That was not necessarily the immediate objective.
Their significance lies in something subtler: they demonstrated that institutions built with China at their centre could attract members, perform functions and coexist with the existing international system.
That experience reduces the uncertainty surrounding the next institutional experiment.
Influence does not always look like a new treaty
There is also a quieter side to this strategy.
China has supported political-training institutions in several African countries with liberation-party histories. CCP-linked schools have provided training in governance and administration, including through a facility reported to have cost around $40 million.
This is very different from creating a development bank.
There is no large capital pool and no formal multilateral treaty to point to. The influence operates through officials, administrative practices and political relationships.
That matters because institutional power is not produced only by treaties.
It can also develop through the people who learn how to operate governments, the networks they build and the administrative models they encounter.
The result is a broader form of institution-building: financial institutions at one level, political and security organisations at another, and networks of officials and governance practices underneath them.
Each operates differently.
But each creates relationships that can outlast the announcement that created them.
Why AI presents a different opportunity
This is where WAICO becomes particularly interesting.
There is no Bretton Woods equivalent for artificial intelligence.
There is no single institution whose authority over global AI governance is already settled. Instead, governments, international organisations and technology powers are developing different approaches to AI safety, standards, access and governance at the same time.
That makes AI unusually open to institutional competition.
In a mature system, a new institution has to work around rules that already exist.
In AI, many of those rules are still being written.
That creates a first-mover opportunity.
A country that helps establish an institution early can influence not only individual policies but also the language, standards and assumptions through which later negotiations take place.
The proposed location matters too. A global AI-governance organisation headquartered in Shanghai would give China something more than another diplomatic platform. It would give Beijing an institutional base from which to participate in defining how the international AI system develops.
That does not mean China would automatically control that system.
It means China would be helping build one of the places where the system is negotiated.
The previous experiments matter because they were experiments
The sequence from AIIB to WAICO is important precisely because none of the earlier institutions was an overwhelming success on its first day.
The AIIB had to demonstrate that countries would actually join and that the institution could function credibly.
BRICS had to demonstrate that a grouping of major emerging economies could expand despite substantial differences among its members.
The SCO had to develop beyond a narrow regional arrangement.
China’s political-training initiatives had to operate without producing a backlash large enough to force them into retreat.
Each successful experience answered a slightly different question:
Would other states participate?
Would they continue participating?
Could a China-backed institution perform a useful function?
Could it acquire legitimacy without replacing the existing system?
The answers were not identical in every case, and none of these institutions is without serious limitations. But their continued existence matters.
By the time China proposed WAICO, Beijing had considerably more experience with institution-building than it had when the AIIB was first proposed.
WAICO raises the stakes
AI governance is different from development finance because there is much less institutional ground already occupied.
That makes the competition more consequential.
If WAICO becomes durable, China would not simply have another organisation through which to discuss AI. It would have a platform for developing norms and standards around a technology whose international rules are still taking shape.
That is a much more ambitious proposition.
And it is already encountering competition.
UN-linked AI governance processes continue to develop alongside China’s proposal. Western governments have expressed reservations about Beijing’s approach. China’s own officials have increasingly criticised Western approaches to AI using the language of “AI hegemonism.”
The emerging picture is therefore not one of a Chinese institution replacing a Western one.
It is more complicated.
Multiple models of AI governance are beginning to develop at the same time, and none has yet secured uncontested authority.
That uncertainty is important. WAICO could eventually become a meaningful institution. It could remain one framework among several. It could converge with existing UN-led processes. Or it could struggle to attract sufficient participation to matter.
There is no basis yet for assuming which outcome will prevail.
The contest is ultimately about who writes the rules
Institutional influence tends to become harder to dislodge once institutions acquire users.
Governments establish bureaucracies around them. Officials develop relationships through them. Companies adjust to their standards. Other institutions begin coordinating with them.
The institution becomes part of the environment rather than merely another organisation inside it.
This is what makes China’s earlier institutional efforts relevant to WAICO.
The AIIB did not replace the World Bank.
BRICS did not replace the G7.
The SCO did not replace the established security architecture.
But they demonstrated that China could help construct institutions that other countries were willing to use.
WAICO takes that experiment into a domain where the stakes are considerably higher.
The question is no longer simply whether China can create another institution.
It is whether it can help establish the institutional vocabulary through which the world eventually governs artificial intelligence.
China does not need to win immediately
This is perhaps the easiest part of the strategy to misunderstand.
WAICO does not have to become the world’s dominant AI institution for its creation to matter.
If it attracts significant members, develops functioning programmes and acquires enough legitimacy to influence international negotiations, China will have achieved something important: a durable institutional foothold in a domain whose global rules are still being written.
That is different from gaining influence inside an existing organisation.
It is influence over the architecture itself.
Whether WAICO ultimately succeeds remains an open question.
But the broader trajectory is easier to see.
China has spent years learning how institutional influence can accumulate: establish an organisation, attract participants, make it useful, allow legitimacy to develop, and then use that experience to attempt something more ambitious.
WAICO brings that process into artificial intelligence.
China is not simply trying to participate in the future rules of AI. It is trying to have a hand in writing them.