Artificial intelligence is advancing at an accelerating pace, and innovation in both AI development and AI governance has become a defining question of the present era, one that bears directly on national competitiveness and on the trajectory of human civilization. In his keynote address at the opening ceremony of the 2026 World AI Conference and High-Level Meeting on Global AI Governance, President Xi Jinping called on the international community to join hands in building a just and equitable system for global AI governance. On 24 July 2026, the Ninth Plenary Session of the 12th CPC Shanghai Municipal Committee reviewed and adopted the Opinions of the CPC Shanghai Municipal Committee on Accelerating the Building of a Hub for AI Development and Governance Innovation (hereafter “the Opinions”). The document directs the city to accelerate the construction of an AI innovation hub of global influence, to contribute to a just and equitable system of global AI governance, and to place itself at the forefront of the world in both the development and the governance of AI. It marks a significant institutional outcome of Shanghai's efforts to implement General Secretary Xi Jinping's remarks during his inspection of the city and his instructions on its work. How Shanghai should locate itself within the national strategic design, draw on its strengths and remain at the forefront, and how it can translate comparative advantage into demonstrable innovation results, are questions that now warrant close study.
I. Why Development and Governance Innovation Are Emphasized in Tandem
The Opinions accord equal weight to AI development and to governance innovation. Several considerations may account for this pairing.
First, AI development is the precondition for AI governance. To discuss governance in the absence of development is to leave that governance without substance. This conclusion follows from the basic principles of historical materialism. In the Marxist account, the forces of production determine the relations of production, and AI governance belongs largely to the domain of the relations of production. Without a corresponding productive base, effective participation in governance is not possible. This is one important reason why we place such heavy emphasis on AI capacity building. The United States enjoys a strong first-mover advantage in AI, and several of its frontier AI organizations, among them Anthropic, OpenAI and Google, have undertaken sustained and consequential work on AI governance. If China is to cooperate with them more effectively on governance, it must first establish a foundation in capability, and above all in frontier model development. China's open-source models have made an important contribution here. Following major breakthroughs by Zhipu AI, Moonshot AI, DeepSeek, MiniMax and others, China has been able to participate in AI governance far more effectively. The catch-up of Chinese open-source models at the frontier has given China and the countries of the Global South the standing to take part in the global dialogue on AI governance.
Second, the arrival of artificial general intelligence has brought governance to a critical juncture. Although the academic community has yet to converge on a definition of AGI, the prevailing positions all point clearly in the same direction, namely that AGI is arriving quickly. The most aggressive forecasts place its arrival somewhere around 2026 to 2027. Even relatively conservative projections expect an accelerating approach within five years. As a disruptive technology, AGI will have highly complex effects across human society. As General Secretary Xi Jinping has put it, we should take seriously the various types of inherent and secondary risks that AI may trigger. The emergence of such risks means that AI governance must move high up the agenda. Getting governance right is, moreover, itself a precondition for the healthy development of AI.
Third, AI development and governance innovation reinforce one another. Anthropic offers a useful illustration. The Anthropic team broke away from OpenAI. When ChatGPT was released in November 2022, OpenAI was the leading research organization in the entire industry; three years on, Anthropic has pulled ahead. The core question is why. The reasons are complex, but one important factor is that Anthropic has consistently taken governance seriously. Its core team left OpenAI precisely because of disagreements with OpenAI founder Sam Altman over questions of governance. Anthropic's early formulation of “Constitutional AI” and of the three-H principles (helpful, honest and harmless), together with its subsequent frontier work on model interpretability and on the question of whether AI systems might develop consciousness, have kept the company at the leading edge of AI governance. Its governance work, and its research on interpretability in particular, has likewise provided important support for making its models feel more recognizably human. Put differently, the Anthropic case partly demonstrates that building influence over the governance agenda, and investing in governance research, can contribute to progress on foundation models themselves. And so it has proved. As the capabilities of foundation models continue to grow, the task of harnessing them, now commonly termed harness engineering, becomes correspondingly more important, and research on governance questions contributes directly to that work.
II. Shanghai's Comparative Advantages
In AI development and governance innovation, Shanghai holds the following comparative advantages.
First, an international advantage. Shanghai continues to build out its roles as an international financial center and an international shipping center, and its level of institutional opening-up has risen steadily. Major international events such as the China International Import Expo and the World AI Conference are held in the city year after year, serving as important windows onto China's openness and as connectors to global innovation resources. In July 2026 the World Artificial Intelligence Cooperation Organization was formally established in Shanghai, with its headquarters in the city. The BRICS New Development Bank is likewise headquartered there. From event platforms to permanent international institutions, Shanghai's international standing is undergoing a structural upgrade. AI development and governance depend on broad international participation and deep collaboration, and Shanghai's distinctive international position allows it to provide solid platform support for both technological innovation and the construction of a global governance system.
Second, a rule-making advantage. Shanghai has been experimenting with rules for AI for some time. The Regulations of Shanghai Municipality on Promoting the Development of the Artificial Intelligence Sector, issued in 2022, were China's first provincial-level local AI legislation, creating both room for experimentation and institutional safeguards for emerging AI business models. In July 2026 the city further released the Guidelines for Building Shanghai's AI Standards System (2026–2028), aimed at providing unified technical standards for AI development.
Third, an industrial advantage.Shanghai has accumulated considerable depth in the AI industry. It has assembled a cluster of leading domestic AI-chip firms, exemplified by Biren Technology and MetaX; a cohort of foundation model companies including MiniMax and StepFun has emerged; and the country's first corpus operations platform has been built there. Shanghai's industry is likewise a domestic leader in embodied intelligence and autonomous driving. The city has also cultivated a distinctive industrial innovation ecosystem, including the Shanghai Foundation Model Innovation Center (“Model Speed Space”) in Xuhui and the Model Magic Community in Pudong, which together have produced an effective physical clustering of the AI industry.
Fourth, a talent advantage. Shanghai has developed a multi-layered system for cultivating and attracting AI talent, organized around high-caliber universities and complemented by a matrix of universities, research institutes and new-model training institutions pursuing differentiated paths. ShanghaiTech University, jointly established and jointly built by the Shanghai Municipal Government and the Chinese Academy of Sciences, is a leading example of the integration of research and education. The Shanghai Innovation Institute has piloted a model that encourages students to “learn through research, innovation and entrepreneurship,” guiding them to tackle hard problems collectively on real projects and assessing them along four dimensions of impact: research, development, industry and society, with entrepreneurship and the creation of tangible value explicitly rewarded. Institutions such as the Shanghai Academy of Natural Sciences are experimenting with close-peer (specialist-peer) review to provide long-horizon, stable support for young scholars pursuing high-risk, high-value basic research, building a talent reserve for foundational AI innovation.
III. From Comparative Advantage to Innovation Outcomes
To turn Shanghai's comparative advantages in AI development and governance innovation into genuine results, work can proceed along the following lines.
First, treat the arrival of the World Artificial Intelligence Cooperation Organization as an opportunity to build out the supporting apparatus. Shanghai can make full use of its international platforms and the institutional advantages of the pilot free trade zone to serve national strategy proactively. It might, for example, establish a dedicated task force on global AI governance cooperation, focused on such key issues as governance capacity-building training, cross-border data flows and the mutual recognition of algorithm filings, and produce workable arrangements aligned with high-standard international rules ahead of others. Drawing on the city's own experience in AI ethics and cross-border data flows, training curricula could be extended beyond purely technical R&D to cover global governance rules, security compliance and industrial cultivation, helping developing countries build up local technical backbones and governance professionals. As international AI application cooperation centers are established with partner countries, Shanghai can serve as a core host zone or major collaborative node, drawing on its AI industry and its abundance of megacity application scenarios to provide technical matchmaking, scenario validation and industrial incubation support to the regional centers.
Second, push institutional innovation in the rules domain. Shanghai can explore a further deepening of pilot-zone mechanisms and pursue institutional breakthroughs, for instance by establishing AI innovation sandboxes in selected districts so that innovation can iterate rapidly within a compliant framework. The city can also build a position of strength and influence in indices and evaluation by publishing AI governance indices, and, leveraging its financial and data-market advantages, make those indices an important reference point for corporate investment and financing decisions and for government procurement.
Third, further cluster and develop advantage industries. By combining broad advance with targeted breakthroughs, industrial agglomeration can be converted into development results. Around the key industrial nodes most likely to break through in the next phase, infrastructure construction can drive an overall breakthrough, for example through an ultra-large-scale world-model simulation training ground supplying low-cost synthetic data and high-fidelity test environments for embodied intelligence and related fields.
Fourth, explore new models for cultivating innovative talent. Beyond the existing talent-training system, priority should go to new pathways suited to the AI era. Support for career transitions also needs attention. Given the employment pressure facing programmers as AI advances, they can be guided toward the emerging role of “forward deployed engineer” (FDE), with corresponding evaluation standards and scaled training pathways established for such new positions.
Fifth, explore new organizational forms. AI represents a wholly new revolution in the forces of production, and it must be matched by new relations of production, which in turn requires a comprehensive transformation of organizational forms. Innovative experiments with the one-person company (OPC) and similar arrangements have already appeared in practice across the country. These experiments have generated considerable enthusiasm among participants, but they remain at a very early stage. How to organize these energetic individuals more effectively, and to furnish them with physical space, capital and technical support, will be the key question in the next phase of exploration.
By Gao Qiqi, Research Fellow, Center for Global AI Innovative Governance (CGAIG); Professor, School of International Relations and Public Affairs, Fudan University
Original Link：https://www.shobserver.cn/staticsg/res/html/web/newsDetail.html?id=1155504&sessionid=
