En
- Research Outputs Think Tank Reports Academic Papers Current Affairs Commentary Research Reference Academic Works
- Academic Exchange International Conferences Domestic Conferences Domestic and International Exchange
- External Engagement Podcasts Newsletters Media Interviews Communication Reference
- Center Team Academic Committee Research Team Special Experts Visiting Scholars Administrative Team
- Talent Development International Youth Exchange Strategic Youth Tsinghua University Student Association for Global Strategy Studies Tsinghua University Student Association for China-US People-to-People Exchange Overseas Practical Activities Doctoral Researchers
Research Outputs
Current Affairs Commentary
Xiao Qian Publishes Article in South China Morning Post Responding to Discussions on Slowing AI Development
2026-10-22
Xiao Qian Deputy Director of the Center for International Security and Strategy, Tsinghua University; Vice Dean of the Institute for AI International Governance, Tsinghua University
Xiao Qian, Deputy Director of the Center for International Security and Strategy at Tsinghua University, recently published an article in the South China Morning Post analyzing the question recently raised by the U.S. side about "whether China agrees to slow the development of frontier artificial intelligence."
The article notes that as frontier AI capabilities continue to break through, discussion is intensifying over whether the pace of AI development should be slowed. Some U.S. tech companies and researchers have called for pacing controls on frontier AI development, and some have even proposed pausing the development of certain capabilities. During the author's visit to the United States, Americans repeatedly raised the following question: Would China agree to slow the development of frontier artificial intelligence?
The article attempts to answer this question from a Chinese perspective, without simply giving a "yes" or "no" answer, and raises several questions that are more worth considering.
First, who can answer on behalf of "China"? Chinese AI safety researchers, model development companies, and the general public have different understandings of AI risks, technological development, and calls for slowing down. It cannot simply be assumed that there is a single unified answer in China.
Second, what exactly should be slowed down? Is it model training, computing power investment, capability enhancement, or the deployment of high-risk capabilities? For Chinese companies that are still working to close the frontier technology gap and face computing power constraints, this is not entirely the same problem as that faced by leading U.S. laboratories.
Third, how can credible and verifiable arrangements be established? If even the boundaries of frontier capabilities, the criteria for judging dangerous capabilities, slowdown measures, and verification mechanisms have not yet been clarified, it is very difficult to expect countries to reach binding agreements. At the same time, the U.S. frontier companies calling for a slowdown also need to prove the credibility of their safety commitments through their own actions.
The article also emphasizes that understanding China's position cannot be separated from its historical development experience and the current China-U.S. science and technology relationship. China attaches great importance to scientific and technological progress and national development, and also emphasizes the balance between development and security. Against the backdrop of the United States' continued technology restrictions on China, how to avoid AI safety cooperation being seen as a tool for maintaining technological leadership is a real problem that must be faced.
The article concludes with three recommendations: China and the United States should continue dialogue on frontier AI risks and benefits; draw on the experience of the Pugwash Conferences on Science and World Affairs during the Cold War to create space for technical experts from both countries to explore risk assessment and verification mechanisms; and encourage Chinese and U.S. companies to take parallel safety assessment and risk prevention measures first, and even engage in healthy competition over "who can develop safer AI."
The article's basic judgment is that rather than rushing to demand that China commit to slowing AI development, it is better to first establish the technical and institutional foundations that would give such a commitment practical meaning. This concerns not only whether China is willing to participate, but also whether future global AI safety governance can truly move toward operable international cooperation.
This article was first published in the South China Morning Post on October 6, 2026
Original link:
https://www.scmp.com/opinion/world-opinion/article/3369561/would-china-agree-slow-ai-development-thats-wrong-question?pgtype=topic
More Recommendations
Zhao Minghao: What Impact Will America's "New Look" Have on China?
Jiang Feng: A Trade War with China Will Not Save Europe
China-Thailand Green Industry Cooperation in the Context of the Global Energy Crisis
Song Qingrun: A New Window of Opportunity for Myanmar-ASEAN Relations?
Substack
Copyright ©2019 Center for International Security and Strategy, Tsinghua University All rights reserved
Email: ciss@mail.tsinghua.edu.cn
Postal Code: 100084
Address: Room 217, Mingzhai, Tsinghua University, Haidian District, Beijing
Room 428, Mingli Building, Tsinghua University, Haidian District, Beijing