# July 17: China Places Its Piece on the AI Governance Chessboard

— As the world enters an era of three parallel rulebooks

In the summer of 1956, a two-month academic workshop was underway at Dartmouth College in Hanover, New Hampshire. A few dozen attendees discussed a question that sounded like science fiction at the time: Can machines think? That gathering was later recognized as the founding moment of artificial intelligence as a field.

Seventy years later, the answer no longer matters as much as it once did—because machines are already making an increasing number of decisions for human beings. From route planning and credit screening to resume filtering and advertising targeting, algorithm-driven choices happen every second, and they no longer feel like science fiction to anyone.

The question that matters has changed. It is no longer "can machines think?" but two harder ones: who sets the rules for these machines? And which set of rules becomes the global standard?

On July 17, 2026, at the World AI Conference (WAIC) in Shanghai—the world's largest AI industry gathering—an answer was laid on the table.

I. Four Opinions, Three Audiences

President Xi Jinping delivered a keynote speech at the opening ceremony of WAIC 2026. The speech was concise in length but clear in structure.

On its surface, the speech laid out four core propositions: open and win-win cooperation, safety and controllability, inclusiveness and diversity, and solidarity and shared responsibility. But when placed in the broader context of current global AI governance dynamics, a more nuanced narrative emerges: each of these four propositions addresses a distinct audience.

Addressing the United States: the "small yard, high fence"

The speech contained a pointed formulation: "jointly oppose the practice of overstretching the concept of national security in AI and placing one's own security above that of others."

The context is unmistakable. Over the past several years, the United States has imposed progressively tightened export controls on advanced AI chips and semiconductor equipment to China—from the initial October 2022 restrictions on high-end AI chips, through the tightening of H100 and H800 supply in 2023 and 2024. The strategy, known in policy circles as "small yard, high fence," effectively expands technology security into national security. China's response, rather than retaliatory measures, has been framed as "open and win-win"—you close your door, we open ours.

The timing is noteworthy. On the same day as WAIC's opening, the U.S. side was also issuing tough signals on tariff pressure and criticism of China (SCMP, 2026-07-17). The "openness" framing of Xi's speech can be read as a direct counterpoint to the other side's messaging on that same day.

Aligning with Europe: the shared language of rules-first

The "safety and controllability" proposition carries a key implicit message: China has rules.

Europe's dominant AI governance anxiety has long been that technology moves faster than regulation. The European Union responded with the AI Act—the world's most comprehensive AI regulatory framework, entering full enforcement in 2026. The Act uses a risk-tiered approach: higher-risk applications face stricter requirements, the highest-risk are outright banned, and all AI-generated content must be labeled. The EU has also established a dedicated AI Office for enforcement.

By contrast, the United States still has no comprehensive federal AI law. Oversight relies primarily on voluntary corporate principles and executive orders, the latter of which can be rescinded by the next president.

Against this backdrop, "safety and controllability" is effectively a signal to Europe: on the "regulate first" position, China stands closer to Europe than to the U.S. A December 2025 Nature editorial made a similar observation, noting that America's regulatory vacuum is eroding its rule-setting credibility in AI (Nature, 2025-12).

Engaging the Global South: building institutional bridges

The third and fourth propositions—"inclusiveness and diversity" and "solidarity and shared responsibility"—address a deeper structural concern: AI must not benefit only a small number of wealthy nations, or the global intelligence divide will only accelerate.

This is not an empty statement. The most substantive parts of the speech follow in the form of three concrete commitments.

II. Three Commitments: Aid, or Infrastructure for Rule Export?

President Xi announced three specific commitments during his speech. The significance of each lies not in the scale of the aid itself, but in the direction they collectively point toward: China is building a global infrastructure for AI governance.

5,000 training slots

Over the next five years, China will provide 5,000 AI training positions for developing countries.

Five thousand slots is not a large number by diplomatic standards. What matters is what these trainees will take away. If their training is grounded in China's own AI governance framework—covering technical standards, ethical principles, and regulatory systems—then when these individuals return to their home countries and participate in AI policy formation, the reference framework that first comes to mind may well be Shanghai's, not Brussels'. This is not technical training; it is a pathway for governance diffusion.

Six regional AI cooperation centers

China will establish International AI Application Cooperation Centers targeting six regional organizations: ASEAN, the Arab League, the African Union, CELAC, the Shanghai Cooperation Organization, and BRICS.

The operational logic is noteworthy. Rather than negotiating with individual countries, China has directly reached the six most significant multilateral groupings covering nearly all of the developing world—Southeast Asia, the Arab world, Africa, Latin America, Central Asia, and the major emerging economies.

This is a platform-style expansion strategy. The centers, based in China, offer technology collaboration, standards co-creation, and solution deployment. For Global South countries, this means a one-stop AI cooperation entry point—they no longer need to navigate multiple international platforms to access both technology and governance frameworks.

The "Mazu" weather system across 30 countries

Mazu is an AI-powered meteorological early warning system, helping countries improve the accuracy of extreme weather forecasts.

Weather was chosen as the entry point for good reason. Meteorological early warning is a high-need, low-political-barrier public good—no country will refuse better typhoon prediction. Through this uncontroversial service scenario, China's AI system gains access to the government infrastructure of 30 countries.

The pattern is familiar: first build the basic service tool, then expand into governance cooperation. It mirrors the playbook of mobile payments—first offering a utility bill payment service, then expanding into financial services once trust and adoption were established.

The cumulative picture

Taken together, these three commitments point toward a single trajectory: China is not simply offering aid packages at WAIC—it is assembling a governance infrastructure that others can adopt. This is a role transition from "technology taker" to "rule maker," achieved not through rhetoric but through concrete channels of training, institutions, and services.

III. Three Rulebook Logics: Parallel Systems of AI Governance

Stepping back from Shanghai and looking at the regulatory regimes of the world's major economies, three fundamentally different logics emerge. They differ not in degree but in underlying philosophy.

The European model: rules first, then development

The EU AI Act is the world's first comprehensive AI law, entering full enforcement in 2026. Its core design is a risk-tiering system that classifies AI applications into four levels, with the highest-risk uses (such as social scoring and real-time facial recognition) banned outright. AI-generated content must be clearly labeled, and violations carry substantial fines. The EU has established a dedicated AI Office for enforcement.

The logic is precautionary: define the boundaries legally, then let businesses operate within them. Its strength is regulatory certainty; its weakness is potential drag on innovation. The Act also produces a "Brussels Effect"—other jurisdictions often model their own AI rules on it.

The American model: run first, regulate later

The United States has the strongest AI technology ecosystem in the world but has moved the slowest on rules. There is no comprehensive federal AI law. Oversight relies on two mechanisms: voluntary AI principles published by major companies (non-binding), and executive orders (reversible by the next president). Federal legislative efforts remain stalled.

This model has been described by some analysts as a "slow-moving giant." Its advantage is minimal constraint on innovation; its disadvantage is a fragmented, uncertain compliance environment as individual states (Colorado, Utah, California) push ahead with their own laws. An AIFOD analysis characterized the U.S. landscape as a "fragmented patchwork."

The Chinese model: state-led, top-down

Since 2022, China has introduced a series of regulations including the Algorithmic Recommendation Management Provisions and the Interim Measures for the Management of Generative AI Services, with an AI law in the legislative pipeline. Core features include mandatory pre-deployment safety assessments for AI products, visible watermarks on AI-generated content, and clear enterprise accountability. This framework has been described as one of the most operationally executable regulatory systems in practice (University of Virginia NSDPI, 2026-04).

At WAIC 2026, this logic was further systematized and began extending outward through the WAICO channel.

The fundamental difference

Each of these three systems represents a distinct governance philosophy: the EU believes in the power of rules, the U.S. in the power of markets, and China in the power of the state. The comparison is not about which is superior, but about understanding a basic fact: on the question of global AI governance, there is no single answer—only parallel choices.

But an influential third-party institution has already made its assessment.

IV. Two Signals: Nature's Endorsement and WAICO's Launch

The Nature editorial

In December 2025, the journal *Nature*—one of the world's most prestigious scientific publications—published an unusual editorial. Its title: *China is leading the world on AI governance: other countries must engage.*

*Nature* rarely publishes geopolitical commentary. When it does, the fact of the editorial itself is a signal. Its core judgment is that China is ahead in building an AI governance framework, and other countries risk missing the window for rule-setting if they stay out of the conversation.

Notably, the editorial also acknowledged China's shortcomings—Chinese AI companies scored lower on the Future of Life Institute's AI Safety Index than their Western counterparts, partly due to insufficient attention to catastrophic misuse risks. This balanced assessment actually enhances the editorial's credibility.

WAICO: from proposal to organization

On July 17, a more concrete action took place in Shanghai.

One year earlier—on July 26, 2025, at WAIC 2025—Premier Li Qiang formally proposed the establishment of a "World Artificial Intelligence Cooperation Organization" (WAICO) on behalf of the Chinese government, with its headquarters tentatively planned for Shanghai (Xinhua, 2025-07-26). On July 16, 2026, the WAICO establishment agreement was signed in Shanghai by Foreign Minister Wang Yi (Xinhua, 2026-07-16). The following day, President Xi announced the organization's formal establishment in his WAIC keynote. Pakistan was among 29 countries that signed as founding members (Guandian/Xinhua, 2026-07-17).

From proposal to agreement in under a year. The pace itself is revealing—WAICO is not a document-level concept but an initiative already in substantive implementation.

WAICO's institutional design has a distinctive feature: it bundles "development rights" and "governance rights" into one package. Traditional AI governance bodies either focus on governance without development (like the EU AI Office) or development without governance. WAICO's logic is: to participate in rule-setting, first participate in development. This approach is naturally attractive to Global South countries—for developing nations in urgent need of AI infrastructure, "help me develop" is a more compelling starting point than "regulate me."

This design also addresses a deeper governance paradox: emphasizing regulation without capability-building only widens the technology gap—and the wider the gap, the harder any global governance framework is to implement. Bundling development and governance into one organization is an attempt to address this problem at its root.

V. Two Worlds on the Same Day

At the main venue in Shanghai's Expo Center on July 17, the exhibition area surpassed 100,000 square meters for the first time. Over 1,100 enterprises were exhibiting, with more than 3,000 products on display—over 300 of them global debuts. More than 1,400 international guests attended, including multiple Turing Award and Nobel laureates. This was the picture on the WAIC stage.

On the same day, the global capital markets were telling a completely different story. Global tech stocks continued to slide; the Philadelphia Semiconductor Index suffered a sharp drop. Over $1 trillion in market value was wiped from the semiconductor sector. Investors were collectively asking the same question: when will the massive AI infrastructure spending of the past two years start generating returns?

These two events happening on the same day are not simply "good news and bad news occurring simultaneously." There is a logical thread connecting them. AI is transitioning from the "storytelling" phase to the "execution" phase. If ChatGPT in 2022 captured the world's imagination about AI's potential, then by 2026, the capital markets are sending a clear message: technological imagination must translate into measurable commercial returns, or the story loses its value.

And the establishment of governance frameworks is an essential part of this transition—without clear and predictable rules, there can be no certainty in commercial pathways.

VI. Rules Are Power

The people in that Dartmouth College meeting room in 1956 were debating whether machines could think.

Seventy years later, the question has changed entirely. The two questions that matter now are: who sets the rules for these machines? And what does this mean for everyone whose life is shaped by AI decisions?

On July 17, 2026, China took a structural step toward answering the first question. Not as a "victory"—this is not an event best framed in terms of winning or losing. The more accurate description is that China is completing a role transition: from a technology follower in AI to a builder and exporter of AI governance frameworks.

Who sets the rules defines how AI operates—and that definition affects everyone who uses AI products.

**References**

1. **Xi Jinping's Keynote Speech at WAIC 2026: "Building a Just and Reasonable Global AI Governance System"** — People's Daily / Xinhua, 2026-07-17
2. **Xi's call for AI cooperation, Trump blasts China over tariffs** — South China Morning Post, 2026-07-17
3. **Global tech stocks fall as AI trade goes into reverse** — Financial Times, 2026-07-17
4. **China is leading the world on AI governance: other countries must engage (Editorial)** — Nature, 2025-12. https://www.nature.com/articles/d41586-025-03972-y
5. **Three Rulebooks, One Race: AI Regulation in the U.S., EU, and China** — Communications of the ACM, 2026
6. **Competing AI strategies for the US and China (Congressional Testimony)** — Kyle Chan, Brookings Institution, 2026-04-16
7. **WAIC 2026 Preparations (Shanghai Municipal Government Press Conference)** — Shanghai Municipal Government / CCTV / cls.cn / Securities Times, 2026-07-07/06-17/18
8. **Develop to Advance Security; Secure to Ensure Development: AI Governance in the PRC** — University of Virginia NSDPI, 2026-04-07
9. **Chinese Government Proposes WAICO, Headquarters Tentatively in Shanghai** — Xinhua, 2025-07-26
10. **WAICO Establishment Agreement Signed in Shanghai** — Xinhua, 2026-07-16