Updated
Updated · startupfortune.com · Jul 21
China Weighs Curbs on Overseas Access to Qwen, Doubao as Qwen Tops 113,000 Derivative Models
Updated
Updated · startupfortune.com · Jul 21

China Weighs Curbs on Overseas Access to Qwen, Doubao as Qwen Tops 113,000 Derivative Models

3 articles · Updated · startupfortune.com · Jul 21

Summary

  • Chinese authorities held talks with Alibaba, ByteDance and Z.ai over possible limits on foreign access to advanced AI models, including unreleased systems, though no rules have been adopted yet.
  • Reuters said the discussions, led by the Ministry of Commerce, covered both closed and open-weight models as Beijing shifts from promoting open releases to treating top models as national-security assets.
  • Qwen already has more than 113,000 derivative models on Hugging Face, while Doubao passed 100 million daily active users and Z.ai's GLM-5.2 is widely distributed through 26 OpenRouter providers.
  • That reach is already visible abroad: GuruFocus said Chinese models accounted for 63% of US companies' token usage on OpenRouter in early July, up from under 10% a year earlier.
  • The move echoes recent US export controls on Anthropic's Claude models and highlights a broader race to fence off frontier AI once globally adopted.

Insights

With thousands of Chinese AI models already copied worldwide, can Beijing truly put its open-source genie back in the bottle?
As US firms cut costs with Chinese AI, are they inadvertently funding their main technological and geopolitical rival?
Is China's dominance in AI hardware manufacturing the real checkmate move in the global technology race?

China Weighs Export Controls on Top AI Models: Implications for Global Innovation, Security, and Markets

Overview

China is considering restricting overseas access to its top AI models, a major shift from its usual practice of releasing open-weight systems that anyone worldwide can use. This open approach has helped Chinese AI companies gain international traction, with some U.S. businesses even adopting these free models to save costs. However, growing concerns over technological sovereignty and geopolitical competition, highlighted by recent incidents like Anthropic’s model withdrawal, are driving China to rethink its strategy. This move is part of a broader pattern of reciprocal trade restrictions and export controls between China and the United States, signaling a new phase in the global AI landscape.

...