On July 17, Xi Jinping delivered the keynote address at the World Artificial Intelligence Conference in Shanghai, marking the first time China’s leader has addressed the event in-person since its launch in 2018. His appearance underscored the growing importance Beijing attaches to shaping the global governance of artificial intelligence (AI). In his speech, Xi emphasized four observations, including the promotion of “true multilateralism,” language Beijing increasingly uses to justify the establishment of a parallel global order under the Chinese Communist Party’s leadership.

In line with China’s ambitions to embed its own institutions into the global AI governance ecosystem, the most consequential announcement of the conference came a day before Xi’s speech, with the formal establishment of the World Artificial Intelligence Cooperation Organization (WAICO), to be headquartered in Shanghai. The new international organization may prove to be China’s most ambitious effort yet to shape the future deployment and governance of AI.

Building on China’s 2023 Global AI Governance Initiative, the idea for a China-led international organization on AI was first proposed by Premier Li Qiang at the 2025 conference and reiterated in the Five-Year Plan (2026-2030) passed in March. WAICO signals China’s shift from articulating principles for AI governance to building permanent institutions through which those principles can be operationalized and normalized.

Precisely because it will be a focal point for entrenching China’s approach to AI, its potential success as a norms-setting venue risks enshrining digital authoritarian approaches to AI governance.

Twenty-nine countries signed the agreement establishing WAICO on July 16. The founding members were largely from Africa and the Middle East (Algeria, Cameroon, Congo, Ethiopia, Kenya, Lesotho, Mozambique, Oman, Senegal, and South Africa) and Asia (Cambodia, China, Indonesia, Kazakhstan, Kyrgyzstan, Laos, Malaysia, Myanmar, Pakistan, Tajikistan, and Uzbekistan). From Europe, the signatories were Belarus, Russia, and Serbia, while in Latin America, Brazil, Cuba, Nicaragua, and Venezuela pledged their endorsement. Many of these countries already participate in China’s Digital Silk Road, Belt and Road Initiative, or broader digital cooperation programs, which have fueled rising digital repression.

In his keynote address, Xi Jinping hailed the establishment of the new organization as “a major move by China to answer the call of the Global South and unite the international community together to vigorously promote AI development and governance.”

This framing reflects a broader continuity in China’s approach to global digital governance. The stated commitment to Global South partnership and implied win-win cooperation in AI is guided by China’s earlier rhetorical framing on “a shared future in cyberspace,” which seeks to normalize “cyber sovereignty “– the notion that the digital domain is a natural extension of national sovereignty and that digital governance should be the sole providence of states, unencumbered by international human rights obligations or accountability to independent stakeholders. WAICO risks extending such approaches into the evolving domain of AI governance.

Institutions matter not simply because of the ideas they espouse, but because of the mechanisms through which they diffuse tools and shape norms. International organizations create influence by convening governments, training officials, developing technical standards, coordinating research, and establishing enduring networks. These activities gradually shape what becomes accepted as international best practice.

In that sense, China does not even need WAICO itself to supplant existing fora. China just needs its new organization to become a significant venue for the collective normalization of China’s approaches to AI governance. Participating member-countries will then actively promote China’s preferred approach at other fora that remain more influential.

Xi’s keynote outlined areas of priority for Beijing. To “support global AI development and advance global AI capacity building,” over the next five years, China pledged to provide 5,000 AI training opportunities for developing countries. China will establish AI cooperation centers with ASEAN, the African Union, the League of Arab States, the Community of Latin American and Caribbean States, the Shanghai Cooperation Organization, and BRICS, and invite 30 countries to use China’s AI-powered meteorological warning system, MAZU. These initiatives are more than development assistance. They create channels through which Chinese technologies, regulatory approaches, and standards also become embedded within partner countries’ AI ecosystems.

This approach also reflects China’s growing emphasis on AI as infrastructure. Rather than exporting only individual AI applications, Beijing increasingly seeks to provide integrated ecosystems of models, services, computing resources, standards, training, and digital public infrastructure.

China’s domestic AI industrial strategy reinforces this international ambition. The latest Five-Year Plan calls for accelerating the adoption of Chinese AI technologies globally.

Pointing to one area of strategic priority for Beijing, Xi also used his speech to reiterate China’s support for open-source AI development. Open-source is often presented as a means of democratizing AI, but it also serves a strategic purpose by encouraging the widespread adoption of Chinese models and tools. It is likely no coincidence that China’s Moonshot AI chose to align the release of its latest open-source model, Kimi K3, with Xi’s remarks. Moonshot AI, one of the so-called six AI Tigers, claims Kimi K3 performs as well as leading models from OpenAI and Anthropic at some key tasks. Arguably, this illustrates how China’s commercial AI ecosystem increasingly complements its diplomatic efforts to expand international influence, which will likely see greater facilitation through WAICO partnership.

None of this guarantees that WAICO will become a dominant institution in global AI governance. It remains more aspirational. Even so, when a government whose domestic AI and digital governance approach has consistently normalized digital authoritarian practices seeks to influence the international institutions that may shape the future technologies that govern our lives, it demands heightened caution.

China presented the World AI Cooperation Organization as an expression of “true multilateralism.” Yet its broader record in digital governance suggests the organization will promote a state-centric model of AI governance — one that places governments, rather than multistakeholder institutions and human rights safeguards, at the center of the emerging international AI order. Such an approach risks dislodging human rights safeguards from emerging technologies and AI norms.

International norms are rarely determined by a single declaration or summit. They emerge through the steady accumulation of institutions, standards, technical cooperation, and professional networks. Watching how this institution evolves will be important.