Exclusive-Beijing is looking at curbing overseas access to China’s top AI models, sources say
By Fanny Potkin
SINGAPORE, July 7 (Reuters) – Chinese officials have held meetings with top technology firms over the past month about restricting overseas access to China’s most advanced artificial intelligence models, three people familiar with the discussions said.
The talks follow a series of steps Beijing has taken to keep homegrown AI at home and underscore that China, like the United States, now treats cutting-edge AI as a critical national asset that must be controlled.
Among the companies present at the meetings were technology giants Alibaba and ByteDance, as well as startup Z.ai. It was stated that the people who were not authorized to speak to the media refused to disclose their identities.
Since the emergence of DeepSeek’s R1 model last year, Chinese AI models have made major advances around the world thanks to their lower costs and increasing capabilities. Any decision by Beijing to limit access to these products could have repercussions on AI markets as costs rise for many businesses.
Increasing Penalties Against Artificial Intelligence Theft Discussed
At meetings led by China’s Ministry of Commerce, participants discussed imposing limits on the most advanced AI models, both closed source and more open versions, according to two of the sources.
Officials are talking about making any leak or theft of proprietary AI technology a crime under China’s strict national security law, one of the sources said.
Authorities have also raised the possibility of implementing new measures to restrict who can fund domestic AI startups, the source added.
The scope of potential restrictions is still being discussed, two sources said, adding that they may only apply to future models. It was not immediately clear when, or even if, these would go into effect.
China’s Ministry of Commerce, which oversees export regulations, and the National Development and Reform Commission, the country’s state planning agency, whose officials also attended meetings, did not respond to requests for comment from Reuters.
Alibaba, ByteDance and Z.ai also did not respond to Reuters’ questions.
All three companies have a range of AI models; Some are closed source, some are open source; This means users can download, run and customize basic systems.
Alibaba’s Qwen and ByteDance’s Doubao are two of the most widely used AI models in China. Z.ai recently caused a stir in Silicon Valley because its GLM-5.2 model’s capabilities are close to leading US offerings but at a fraction of the cost.
Artificial Intelligence Models and National Security Concerns
U.S. President Donald Trump’s administration is also deeply concerned about the national security implications of AI, particularly the potential for misuse of American AI products by military intelligence in China, Russia, and other countries of concern.
In June, he ordered that foreign nationals be blocked from accessing Anthropic’s most advanced Fable and Mythos models, prompting the company to disable the models for all users worldwide because it could not verify nationality in real time.
Fable’s export controls, which were designed for the public, were lifted following the introduction of new measures. But Mythos, designed for cybersecurity professionals, is still only available to some “trusted” US organizations.
Some US artificial intelligence experts also said that the US should regulate the use of Chinese artificial intelligence models.
CONCERN ABOUT THE MYTHOS THREAT IN CHINA
Chinese officials are deeply concerned about Mythos’ potential to exploit software vulnerabilities and that Washington could apply the model against Chinese interests, according to two of the sources.
This echoes concerns publicly voiced by state media and Zhou Hongyi, the founder of cybersecurity firm 360, a major supplier to government and corporate clients, who says China needs to develop its own Mythos.
China has implemented numerous measures to protect domestic AI this year.
In April, the country’s state planner ordered Meta to drop its $2 billion acquisition of China-based artificial intelligence startup Manus. In early June, authorities issued sweeping new rules tightening control of overseas deals involving Chinese investors, technology, data and national security.
This year, China also launched an investigation into Manus and other local AI startups that moved abroad to determine whether they violated export control laws, according to two of the sources and a third person.
Manus did not respond to requests for comment.
Reuters was unable to learn how possible new restrictions on access to Chinese AI models from abroad might work.
But some clues can be gleaned from a May roundtable of Chinese legal experts on regulations governing open-source artificial intelligence.
According to a summary of the discussions published in an official Supreme People’s Court journal, participants proposed a tiered system: basic open source tools subject to simple filing, more advanced technologies that face security reviews, and the most sensitive border models banned from public release or restricted to domestic use.
(Reporting by Fanny Potkin in Singapore; Editing by Anne Marie Roantree and Edwina Gibbs)
