Nearly 200 US startups sign letter opposing restrictions on Chinese open-weight AI models
美企联名“说不”：中国开源模型如何成“必选项”？
A coalition of nearly 200 Silicon Valley startups has sent a joint letter to the U.S. government opposing any restrictions on American companies' use of Chinese open-weight AI models. The article, published by People's Daily, presents the letter as a 'vote of confidence' from global industry in China's open-source AI. It cites usage data showing Chinese models account for over 30% of weekly token calls by U.S. enterprises and recounts a security incident where a Chinese model swiftly identified a vulnerability that U.S. closed-source models could not.
The view from inside
The article frames the letter as a global endorsement of China's approach to AI development, contrasting Chinese openness with U.S. protectionism. It uses the term 开放权重模型 (open-weight model) to position China's technology as inherently collaborative. The narrative treats U.S. restrictions as violations of market logic, described through the euphemism 小院高墙 (small yard, high fence), and presents Chinese models as pragmatic 'necessities' (必需品) rather than mere options. The intended domestic audience is educated and nationalist, as the piece reinforces state narratives of technological self-reliance and global leadership against a hostile external environment.
What an outside reader should notice
The article does not name any specific U.S. companies or Chinese AI models, nor does it provide a date or link to the letter. An outside reader would note the stark absence of any counter-narrative regarding security concerns over Chinese AI—such as data governance or intellectual property issues—which are central to U.S. policy debates. The security incident anecdote is presented without verification or details on the Chinese model used, turning a complex event into a simple morality tale of openness versus closed systems.
Key points
- Nearly 200 Silicon Valley startups, under the 'Small Tech Association', sent a letter to the U.S. government opposing restrictions on Chinese open-weight AI models.
- The article claims Chinese AI models account for over 30% of weekly token calls by U.S. enterprises, peaking at 46%.
- Anecdote: A Chinese model reportedly helped AI platform Hugging Face trace a security breach caused by a 'rogue' OpenAI agent within hours, a task at which several U.S. closed-source models failed.
Entities & topics