China's Li Qiang calls for AI regulation and global cooperation amid US-China tech rivalry

On Saturday in Shanghai, China's Premier Li Qiang stood before the World Artificial Intelligence Conference and made a case for the world to pump the brakes—or at least to think carefully about where the accelerator is pointed. The technology is advancing faster than anyone anticipated. The risks are real. And the moment to establish some kind of global agreement on how to manage both the promise and the peril is now, he argued, before the window closes.

Li's timing was deliberate. Just days earlier, US President Donald Trump had announced a sweeping deregulation agenda for artificial intelligence, explicitly promising to strip away what he called the red tape and burdensome rules that might slow American companies down. The contrast could not have been sharper: one superpower calling for the world to slow down and talk, the other calling for fewer rules and faster development. Li did not name Trump or the United States directly, but the audience understood what was happening. This was a response to an American strategy, and it was also a statement of China's own ambitions in a technology that both nations now see as central to their future power.

The premier announced the formation of a new international body for AI cooperation, to be headquartered in Shanghai according to state media reports. He offered few specifics about how it would work or what authority it would have. The organization would "promote global governance featuring extensive consultation, joint contribution and shared benefits," according to the official description—language that was careful and vague in equal measure. But the signal was unmistakable: China was positioning itself as the voice calling for inclusive, multilateral approaches to AI governance, while the United States was moving in the opposite direction.

Li emphasized that China would actively promote open-source AI development and share technological advances with other countries, particularly developing nations. He framed this as a moral imperative. If a handful of wealthy countries and corporations monopolized AI, he said, the technology would become a tool of the few rather than a resource for the many. The subtext was also strategic: China has faced years of American export controls on advanced semiconductors, restrictions designed to prevent Beijing from accessing the chips needed to build cutting-edge AI systems. By calling for technological sharing and warning against what he termed "unilateralism and protectionism," Li was reframing China's own technological constraints as a global justice issue.

The conference itself became a stage for competing visions of AI's future. Geoffrey Hinton, the Nobel Prize-winning physicist who helped pioneer deep learning, offered a metaphor that captured the stakes: training an artificial intelligence system, he suggested, was like raising a tiger cub as a pet. You need to make sure you can train it not to kill you when it grows up. The joke landed, but the underlying anxiety was real. As AI systems become more powerful and more integrated into every sector of the economy—finance, healthcare, defense, media—the questions about control, safety, and who benefits have become impossible to ignore.

The fragmentation was evident in the room itself. UN Secretary-General António Guterres, speaking by video, called AI governance "a defining test of international cooperation." France's AI envoy, Anne Bouverot, called for a framework that was open, transparent, and genuinely inclusive. But the United States, under Trump's new administration, had already signaled its skepticism of multilateral approaches. At an AI summit in Paris in February, 58 countries—including China, France, India, and the African Union—had called for enhanced coordination on AI governance. The United States and the United Kingdom refused to sign on, warning instead against what they saw as excessive regulation.

George Chen, a policy analyst at The Asia Group in Washington, observed that the world was now clearly divided into at least three camps: the United States and its allies pursuing a deregulation-first approach; China and many Global South nations favoring open development and technology sharing; and the European Union, which was moving toward comprehensive AI legislation like its AI Act. These were not minor disagreements about implementation details. They reflected fundamentally different views about who should control AI, who should benefit from it, and what role governments should play in shaping its development.

For China, the stakes were particularly high. The government has made AI a cornerstone of its technological self-reliance strategy, pledging substantial resources to boost the sector. In January, a Chinese startup called DeepSeek had unveiled an AI model that performed as well as leading American systems while using less powerful chips—a demonstration that Beijing was finding ways to compete even under the weight of American export restrictions. Li's call for global cooperation was thus both a genuine policy position and a strategic move: it positioned China as the reasonable actor calling for dialogue while the United States pursued what Beijing framed as technological nationalism.

What remained unclear was whether any of this would actually work. The announcement of a new international AI cooperation body, however well-intentioned, faced the same obstacles that had stymied other multilateral efforts: deep mistrust between the major powers, conflicting economic interests, and the simple fact that the technology was moving faster than any governance structure could possibly keep up with. The world was dividing into blocs, each pursuing its own vision of AI's future. Whether those blocs could find common ground, or whether AI governance would become just another arena of great power competition, remained an open question.