China Proposes Export Controls on AI Technologies for National Security
China has proposed new export controls on several of its domestically developed AI technologies, including models, agents, and computing clusters, according to reports. This initiative may contradict the open-weight design of China's most advanced AI releases but is expected to maintain API access for individual users, allowing these tools to continue gaining popularity internationally.
In recent years, Western developers and the U.S. government have raised concerns about Chinese developers allegedly copying their technology. This has led to various measures aimed at constraining Chinese AI development, including restrictions on access to advanced processors and AI chips. Ironically, China is now looking to limit U.S. and international access to its own AI tools and frameworks.
Citing national security concerns, China plans to impose export controls on 23 key AI technologies. While this could make the affected models less open than initially intended, the continued availability of API access is likely to support the ongoing growth of these models.
Chinese AI models have seen a significant surge in popularity recently. For instance, models like DeepSeek V4.1 Flash and Step 5 Preview have processed over 50 trillion tokens, while the Claude Opus model has only managed 5.85 trillion tokens.
New Measures to Safeguard AI Technology
According to China's commerce ministry, the new export control proposals aim to mitigate the risk of misuse of AI technologies, particularly those capable of generating and deploying autonomous malware that could target critical civilian infrastructure. However, existing open-source licenses are expected to be respected, including those for AI models and academic papers.
Cleared to File
Despite the competitive landscape between the U.S. and China in the AI sector, there have been some collaborative developments in recent months. Following a series of security breaches involving autonomous AI agents, both President Trump and Chinese Premier Xi Jinping agreed during a meeting in September that human oversight should always be a part of nuclear weapons management. Additionally, there was a proposal for an AI hotline-style back-channel between the two nations to provide early warnings and context in case of rogue AI actions.
Other technologies included in these new proposals are solar power and biotechnologies, such as cell and gene therapies. Under these controls, Chinese companies would be permitted to license these technologies but would not be allowed to sell the intellectual property overseas.
