US-China AI rivalry needn’t preclude cooperation

Artificial intelligence (AI) poses risks that do not respect national borders, from cyberattacks and misinformation to system malfunction and potentially dangerous misuse. In a geopolitically charged world, does greater incompatibility between countries make AI safer? UN Secretary General Antonio Guterres offered a clear answer at the inaugural UN Global Dialogue on AI Governance: “When countries align on how to test systems, measure risk and assign responsibility, safety travels with the technology.” By contrast, “a patchwork of incompatible rules raises costs, divides the world – and protects no one”. Yet just as AI creates an unprecedented need for cooperation, geopolitics threatens to pull the world apart. China established the World AI Cooperation Organization (Waico) last month with 29 founding members. Besides AI cooperation and governance, it aims to help developing countries bridge the AI divide. The United States, meanwhile, has developed the Pax Silica initiative around AI, technology and supply-chain security. Both initiatives overlap but do not primarily seek to address the same problems. Waico can promote development and cooperation while strengthening China’s influence in global AI governance; Pax Silica can address genuine concerns about critical technologies and supply-chain resilience while consolidating a US-centred network of trusted partners. Because these initiatives come from fierce competitors, an analytical problem arises when geopolitical rivalry is treated as sufficient to explain their entire meaning. Indonesia joined Waico while engaging in Pax Silica discussions; Kazakhstan joined both. The US is reportedly preparing to tell countries that participation in initiatives deemed incompatible with the American framework could lead to exclusion. Must international AI cooperation reproduce geopolitical blocs? Conflict between countries arises from commercial, strategic and political interests, but perceptions matter too. Once a country is seen primarily as a strategic adversary, cooperation may be read as bloc-building, regulation as political control and ambiguous actions as evidence of hostile intent. This is not merely theoretical. A Carnegie Endowment for International Peace paper warns against dismissing Chinese AI regulations because of the Communist Party being in power or viewing them primarily through geopolitical competition. The picture is complicated: political and information control matter, but policymakers are also responding to AI’s social, economic and ethical consequences. Geopolitical differences can coexist with areas of policy convergence. A 2025 study of US and Chinese AI governance examining more than 40 policy and corporate governance documents found overlap including in system reliability, algorithmic transparency and the use of AI to enhance safety. The analytical mistake lies not in identifying political motives but in treating political motives as exhaustive explanations. Consider human control. At the World AI Conference in Shanghai, Chinese President Xi Jinping called for AI to remain under human control while discussing risk prevention and opposing abuse. Less than two weeks earlier, the UN Global Dialogue on AI Governance included “robust human oversight” among the governance priorities. Yet the same language can be interpreted very differently: Xi’s call, for example, has been read as referring to control exercised by Communist Party institutions. This resemblance in language does not necessarily mean the Chinese and UN approaches have the same motivations, objectives or implementation mechanisms. But neither does institutional difference prove that their underlying policy concerns are fundamentally different. We need to ask what problems a policy seeks to address and how it functions within its institutional environment. A country’s political system should be understood through its history and culture, and form part of the context through which a policy is interpreted – not judged against assumptions about what the system ought to be, with those assumptions used to predetermine what a policy must mean. Communist Party leadership is a central feature of China’s institutional environment. But recognising its importance is different from using it as an all-purpose explanation. An AI policy may reflect political considerations, technological safety, economic development, national security – or several objectives simultaneously. Journalists, think tanks and academics have a responsibility: neither to minimise genuine differences nor manufacture common ground, but to identify both accurately and distinguish evidence from inference. The choices of Indonesia and Kazakhstan suggest another possibility: participation without alignment. Countries can engage with different technological ecosystems and retain room for independent policy choices. Forcing them to choose may harden precisely the divisions global AI governance needs to bridge. The Chinese expression “he er bu tong” means harmony without uniformity. Countries shaped by different histories, cultures and circumstances need not share political systems, values or strategic interests to cooperate. AI will not wait for geopolitical rivalries to subside. If AI is to serve humanity rather than deepen geopolitical divisions, governments must cooperate across differences. Those who analyse international affairs have a role too: not to overlook disagreements, but to take care not to enlarge them through assumptions – and to recognise common ground where it exists. The views expressed here are the author’s own.

US-China AI rivalry needn’t preclude cooperation

.css-1c6uqr6{color:inherit;font-weight:inherit;font-size:inherit;font-family:inherit;line-height:inherit;overflow-wrap:break-word;}Forcing countries to choose between the US’ and China’s initiatives would be counterproductive to global AI governance

Forcing countries to choose between the US’ and China’s initiatives would be counterproductive to global AI governance