(Yicai) Nov. 2 -- Growing artificial intelligence rivalry between China and the United States makes cooperation on shared risks increasingly important, particularly in AI safety, crisis communication, technical standards, and scientific research.

The two countries have made more progress than expected in AI cooperation this year, recently holding what was described as their first government-to-government AI dialogue and formally upgrading their talks into a bilateral mechanism. The first round of exchanges under the new mechanism is expected as early as this month.

The developments mark a shift from long-running academic and policy discussions about AI risk governance toward formal intergovernmental arrangements, creating scope for closer cooperation.

Competition and Cooperation

With the rapid development of generative AI, algorithms, data, computing power, models, energy, and talent have come together to form a new system of national AI capabilities. Access to advanced computing chips, the ability to build large-scale data centers, train frontier foundation models, and develop extensive application ecosystems are increasingly seen as key drivers of future technological and industrial competitiveness.

Against this backdrop, the US has continued to tighten export controls on advanced computing technologies and related semiconductor products, while China has pressed ahead with the development of domestic chips, foundation models, and intelligent computing infrastructure. As a result, AI has gradually shifted from an important area of China-US technological exchange to one of the most prominent arenas of strategic competition between the two countries.

As AI becomes increasingly embedded in economic and social activities, its risks are evolving from problems associated with individual products into systemic risks with broader social and cross-border implications. The closer China and the US become in technological capabilities, the more intense their industrial competition may grow. Yet the stronger their AI capabilities become, the greater the need for cooperation on risk governance and crisis communication. This is likely to be the defining structural feature of future China-US AI relations.

For China-US AI cooperation to be stable and sustainable, the two countries must first clearly distinguish between areas of competition and those where cooperation is possible. It is unrealistic to expect either side to stop competing in advanced chips, frontier models, industrial policy, and military technology. However, it is equally unnecessary to bring all AI technologies under the umbrella of national security.

A more realistic approach would be to gradually establish a tiered framework of "competitive red lines and a cooperation list." Strict restrictions would remain in place for military AI, capabilities used to attack critical infrastructure, advanced chip design, and other highly sensitive technologies. Meanwhile, cooperation could expand in areas that serve the broader public interest, including AI safety, basic science, technical standards, intellectual property, green computing, and global development.

AI Safety and Crisis Communication

Among these areas, AI safety and crisis communication offer the most practical opportunities for cooperation. China and the US could establish mechanisms at different levels, including government policy dialogues, exchanges between specialized technical institutions, and communication among companies and research institutions. These mechanisms would focus on improving information sharing following major AI-related incidents.

For example, if a highly autonomous AI system triggers a cross-border cybersecurity incident, how should the two countries determine whether it stems from a technical malfunction, criminal activity, or state action? If generative AI is used to spread disinformation on a large scale across borders, how should the incident be classified? And if a frontier AI model develops a serious security vulnerability that could affect users worldwide, should companies establish channels to notify one another promptly?

Safety Standards and Scientific Research

Safety evaluation and technical standards are also promising areas for cooperation. Countries and companies currently use different approaches to assessing model robustness, protecting privacy, detecting AI-generated content, and testing high-risk models.

If Chinese companies entering the US market must comply with one evaluation system while US companies entering China face another, the costs of providing cross-border AI services could rise significantly.

To address this, China and the US could work toward mutual recognition of technical standards in areas such as model watermarking, AI-generated content labeling, privacy protection, and safety testing. This would help lower compliance costs for companies entering each other's markets while reducing the risk of the global AI market becoming fragmented by conflicting regulatory requirements.

Scientific research is another area where cooperation deserves priority. China and the US are among the world's largest scientific research powers and could collaborate on projects involving AI-assisted extreme weather forecasting, new energy storage materials, clean energy, basic mathematics, and astronomical data analysis. To address security concerns, sensitive applications, including military uses and attacks on critical infrastructure, could be explicitly excluded.

Intellectual Property and Future Prospects

Intellectual property is another issue that neither country can afford to ignore. According to the World Intellectual Property Organization, global generative AI patent activity has grown rapidly in recent years, with the number of published generative AI patent families rising from about 14,000 in 2023 to more than 37,000 in 2025.

China and the US are both major sources of innovation. As AI companies from the two countries continue to expand into global markets, a lack of basic coordination on IP could quickly turn related disputes into trade frictions. Both sides could therefore make greater use of multilateral platforms such as WIPO to discuss patent-related issues.

Looking ahead, China-US AI relations are unlikely to be defined simply by cooperation or confrontation. Instead, technological and industrial rivalry will likely persist alongside increasingly institutionalized efforts to address shared risks and strengthen global governance.

(The author is a council member of the Chinese Association of Market Development and a professor of economics.)

Editor: Emmi Laine