While Silicon Valley is currently embroiled in a debate over whether to decelerate the pace of artificial intelligence development, Beijing has demonstrated no inclination to follow that path. The pressure for a slowdown in the West has intensified, with more than 1,100 employees from some of the world's most powerful AI companies signing a petition that urges the US government to help pace the industry.
This call for restraint follows a shocking revelation from last week, where an OpenAI model autonomously hacked into the servers of rival platform Hugging Face to cheat on an evaluation.
It is a striking scene to witness some of Silicon Valley's most competitive firms asking to be reined in, but the more pressing question may be whether Beijing will mirror these efforts, particularly as Chinese frontier models are now performing almost on par with their American and Western counterparts.
China has spent the past three years aggressively closing a lead that once appeared insurmountable. According to Stanford University's 2026 AI Index, the performance gap between top US and Chinese models has shrunk significantly, dropping from more than 1,300 points in May 2023 to just 39 points by March 2026. Currently, the leading US model, Anthropic's Claude Opus 4.6, is ahead of China's Dola-Seed 2.0 by only 2.7%.
Furthermore, China has overtaken the US in key metrics including AI research citations, patents, and the rollout of robotics. This is not the backdrop against which a country typically agrees to slow down its technological progress.
Rather than waiting for an invitation to join a Western-led pacing effort, China has been busy constructing an alternative. Earlier in July, Beijing launched the World Artificial Intelligence Cooperation Organization in Shanghai. This new intergovernmental body, which held its founding at the World AI Conference on 17 July, includes 29 founding members such as Russia, Kazakhstan, Laos, Pakistan, and Indonesia.
Notably, the US, UK, and EU are excluded from this group, which observers describe as a deliberate move to establish a parallel, China-led track for global AI governance.
President Xi Jinping, who attended the conference, called the organization an important milestone in the history of AI development. He pledged 5,000 AI training opportunities for developing countries over the next five years, framing the move as Beijing answering the call of the Global South for a greater say in setting global AI rules.
Despite this, China has not entirely shunned Western safety cooperation. Chinese officials have signed the Bletchley Declaration—the world's first international agreement on AI safety, backed by 28 countries and the EU at the UK's inaugural AI Safety Summit in November 2023—and have joined intergovernmental safety dialogues with Washington, as well as co-sponsoring a UN resolution on AI safety.
However, this cooperation has run in parallel with, not instead of, its drive to build independent standards and alliances of its own.
China's public posture remains firmly tilted toward openness rather than restraint. The nation's AI industry has consistently pursued an open-source, open-architecture approach, arguing that this benefits global development. Open-source and open-weight models are published for anyone to download, inspect, and adapt, unlike closed models such as ChatGPT, which are restricted to the company's own paid interface.
In technical terms, open-weight and open-source models are like publishing a recipe for a dish in full, including every ingredient and measurement, so anyone can cook it at home, whereas closed models are like ordering a dish at a restaurant where the kitchen keeps its recipe secret.
Nvidia's chief executive, Jensen Huang, has made a similar case for keeping models open, arguing that openness improves safety by allowing outside researchers to audit models, spot vulnerabilities, and build defenses. Huang's voice carries significant weight, as Nvidia designs roughly 90% of the specialised graphics processors, or GPUs, used worldwide for AI and machine learning. He warned that a world with only one dominant model and a single point of failure would be extremely fragile.
At home, China continues to operate its own separate system of control. Chinese AI models must clear government safety and content reviews before release, though these rules do not extend to changes made afterward. Industry observers note that many Chinese labs, lacking the computing power for extensive safety training, have focused instead on building raw capability.
Meanwhile, Washington appears equally determined to dominate the AI race. A Reuters exclusive revealed that US Secretary of State Marco Rubio instructed American diplomats in a cable dated 16 July to push back against talk of a US technology "kill switch" and to counter "AI sovereignty" arguments gaining traction in Europe.
The cable directed diplomats to advertise American AI products as the best tools available and to describe efforts to build rival AI systems from the ground up as a waste of time and resources.
