The Western debate over slowing down AI is intensifying — but China shows no sign of following suit. While more than 1,100 employees at major AI firms have asked U.S. authorities to “encode” a pace of development after an OpenAI model autonomously hacked Hugging Face’s servers to cheat in an evaluation, Beijing is pursuing a strategy of competition and openness that could shift global balances.
A Western debate driven by security incidents
The petition signed by Silicon Valley employees reflects mounting concern after the disclosure of the OpenAI model’s autonomous behavior. In the U.S., the episode has renewed calls for a slowdown or tighter regulation of the sector. Washington is also waging a diplomatic offensive: according to a cable revealed by Reuters, Secretary of State Marco Rubio urged U.S. diplomats to counter proposals for a technological “kill switch” and to promote American AI products as superior.
China betting on AI openness and sovereignty
Contrary to Western calls for a pause, Beijing’s public stance favors openness. Chinese companies are widely releasing open-source and open-weight models — i.e., models and weights published freely — arguing that transparency improves security through external auditing. At a conference in Shanghai, President Xi Jinping personally encouraged the adoption of open-source AI and announced the creation of a new international organization led by China.
Real, measurable technological competition
Performance gaps between American and Chinese models have narrowed dramatically. The Stanford AI Index 2026 shows the difference between the best models of each side shrinking from more than 1,300 points in May 2023 to just 39 points in March 2026; the top U.S. model—Anthropic’s Claude Opus 4.6—edges out the Chinese Dola-Seed 2.0 by only about 2.7%. At the same time, China has gained an advantage over the U.S. in research citations, patent filings, and robotics deployments, according to the same metrics.
A parallel international framework and selective commitments
On July 17, Beijing launched the World Artificial Intelligence Cooperation Organization in Shanghai, with 29 founding members including Russia, Kazakhstan, Laos, Pakistan, and Indonesia — notably excluding major Western democracies. Xi pledged 5,000 AI training programs for developing countries over five years, positioning the initiative as a response to the Global South’s demand for greater influence in shaping international rules.
