China turns AI governance into a Global South power issue

What this is about

Xi Jinping used the World AI Conference in Shanghai on July 17, 2026, to send a clear message: China does not just want a seat at the global AI governance table; it wants to build a table of its own. In his speech, he argued for a fairer international AI order and put the World Artificial Intelligence Cooperation Organization at the center of that pitch.

This is not ordinary conference language. The speech ties open models, export controls, development support, and technical standards into one political package. For Europe, that matters because AI rules are no longer made only in Brussels, Washington, or individual company labs. They are becoming part of a global contest over influence.

What WAICO actually does

WAICO is meant to serve as an international cooperation structure for AI. According to the published speech, China wants to involve Global South countries more deeply in AI development, governance, and capacity building. Xi announced 5,000 training and seminar opportunities for developing countries over five years. He also said China would expand cooperation with ASEAN, the League of Arab States, the African Union, CELAC, the Shanghai Cooperation Organization, and BRICS.

Another concrete element is MAZU, an AI-based meteorological warning system that China says it will make available to 30 countries. The message is clear: China is presenting AI not only as a software business, but as an infrastructure offer. Whoever supplies compute, models, warning systems, and training also shapes technical dependencies and political standards.

Why it matters

The United States and Europe often talk about safety, export controls, copyright, and liability. China is leaning more strongly into access, open ecosystems, and cooperation with countries that have less AI infrastructure of their own. That is attractive for governments that do not want to wait until Western providers offer services that are cheap, locally adapted, and politically neutral.

At the same time, it is unclear how independent such an organization would be in practice. The official speech stresses the role of the United Nations and real multilateral cooperation. AP and other reporting also frame the move as a counterweight to Western-led AI rules. That is the tension: broader access to AI can create real benefits, but it can also create new dependencies.

For companies and developers, the story is practical. Anyone deploying AI products internationally must understand more than the EU AI Act or US rules. In many markets, Chinese reference architectures, open model families, and state-backed programs may become the working baseline faster than Western compliance papers.

In plain language

WAICO is like a new international toolbox. When people build a house, they often use the tools that are available, affordable, and easy to learn. If China gives many countries that toolbox early, those countries also learn China’s measuring tape, screws, and building plans. Later, switching to a different system becomes harder.

A practical example

Imagine a small coastal country with 12 million people. It faces 40 severe weather events a year, but has only a small national data center and a limited pool of AI specialists. If MAZU gives that country better storm warnings, and 50 civil servants and engineers are trained through Chinese programs, the immediate result could be lives saved.

After two years, though, local data pipelines, training plans, and procurement decisions may also be tied to that ecosystem. A European provider that later offers an alternative warning system is then competing not just with software, but with an institutional network that is already in place.

Scope and limits

First, it is not yet clear how binding WAICO will be, how it will be funded, or which states will take active roles. An announcement is not the same as a durable governance structure.

Second, open AI cooperation can be genuinely useful, especially for countries without frontier infrastructure of their own. At the same time, training programs, model access, and technical standards can deepen political dependencies.

Third, the safety claim is hard to verify. The speech says AI should remain under human control. Whether open models, state programs, and geopolitical competition actually produce more control remains uncertain.

SEO & GEO keywords

WAICO, World Artificial Intelligence Cooperation Organization, World AI Conference 2026, Xi Jinping AI, China AI governance, Global South, AI regulation, open AI models, MAZU warning system, BRICS AI, ASEAN AI, global AI standards