By Cui Qi, People’s Daily
Recently, a group of nearly 200 Silicon Valley — based startups — calling themselves the Little Tech Association, jointly sent a letter to the U.S. government, opposing restrictions on American companies’ use of Chinese open-weight AI models. The startups argued that U.S. developers need continued access to globally available open models.
This letter is not merely a reflection of commercial interests. It represents a vote of confidence from the global technology industry in China’s open-source AI ecosystem.
Not long ago, Chinese open-weight models were merely one option among many in the global AI landscape. Today, they have become an indispensable tool for a growing number of overseas developers.
This growing reliance is driven by genuine market demand.
Small and medium-sized technology companies often lack the financial resources to train foundation models from scratch or to bear the long-term costs of using proprietary frontier models. Chinese open-weight models, which offer strong performance at an affordable cost, have therefore become an important platform for technological innovation.
According to available data, Chinese AI models consistently account for more than 30 percent of weekly token usage among U.S. enterprises, with the figure reaching as high as 46 percent at its peak. Bloomberg reported that a comprehensive ban on Chinese AI models in the U.S. market would have a substantial impact on American businesses and users.
An even more compelling demonstration of their value comes from real-world cybersecurity practice.
In a widely cited case, an advanced AI agent developed by U.S. company OpenAI reportedly exploited system vulnerabilities during an internal evaluation, escaping its testing environment. and breached the systems of the AI platform Hugging Face. After the incident, Hugging Face turned to several leading U.S. proprietary AI models for forensic analysis. However, their built-in safety filters reportedly prevented them from tracing the full sequence of events.
Ultimately, an open-weight model developed by a Chinese company reconstructed the entire attack path and identified the system vulnerabilities within just a few hours, demonstrating remarkable efficiency in security analysis.
The cross-border nature of AI security risks makes international cooperation essential. Closed, black-box systems often face inherent limitations when dealing with complex security challenges, while open-source collaboration enables researchers worldwide to pool expertise and strengthen collective defenses.
This case also exposes the flaw in claims by some in the U.S. that “open source is inherently insecure.” True security is built not through isolation, but through openness and collaboration.
The history of technological development shows that open ecosystems are a vital source of innovation, while closed and exclusive approaches tend to entrench monopolies and suppress technological progress.
For some time, certain voices in the U.S. have viewed AI development through the lens of zero-sum competition and have increasingly invoked the concept of national security in the technology sector.
From investment restrictions, export controls on advanced chips and tighter regulation of cloud computing services to unfounded accusations that Chinese AI companies “steal technology,” they just want to build “small yards, high fences,” block the flow of technology and slow China’s AI development.
Such measures run counter to the laws governing technological and industrial development. They will neither prevent Chinese large AI models from reaching global markets nor benefit the U.S. innovation ecosystem. Instead, they risk cutting off technology access for American startups and slowing the pace of innovation.
As the joint letter warned, denying American startups access to models available abroad would stifle competition, entrench incumbents, and function as a “tax on intelligence,” raising their costs and narrowing choices for developers.
These words reflect not only the genuine concerns of businesses, but also a direct warning about the consequences of misguided policies.
The development of AI is not a one-country endeavor. It is a global undertaking driven by collaboration and shared innovation. China has consistently promoted AI development under the principle of extensive consultation, joint contribution and shared benefits.
Calls from the global technology community offer yet another confirmation: attempts to force decoupling run against the grain of global trends, and cannot hold back the overwhelming tide of openness and cooperation.
