Beijing wants to curb overseas access to China's leading AI models: Here's why

Chinese authorities have convened meetings with prominent tech firms over the past month about potentially barring foreign access to China’s most advanced AI models.

The current talks follow a number of significant steps by Beijing to keep domestic AI within the country and underline how China like the US is now treating groundbreaking artificial intelligence as a strategic asset that needs controls.

Since the inception of DeepSeek’s R1 model last year, Chinese AI models have expanded internationally due to their low costs and increasing capabilities.

Any sort of decision by Beijing to restrict access to those products could have a knock-on effect across AI markets as costs for many businesses would likely increase.

According to officials, using AI technology could constitute an offense under China’s strict national security legislation. They further raised the feasibility of enacting new policies to restrict who can fund domestic AI startups.

The scope of the possible rules is still under discussion, and they may only apply to future models. The Trump administration has also been deeply concerned about the national security implications of AI, specifically the potential for American AI products to be misused by military intelligence in China, Russia and other countries.

Anthropic has earlier announced on Friday it is restricting its powerful new AI model, Fable 5 and Mythos 5, to adhere to export regulations after US authorities identified flaws within the system following its public release.

The company has barred foreign nationals from using Claude Fable 5- a newly introduced action that the company described as “too powerful” for public release.

Chinese authorities are concerned about the potential for Mythos to exploit software vulnerabilities. Consequently, China has also launched investigations into Manus and other local AI startups, attempting to establish whether they have broken export control laws. Additionally, more advanced technologies face security reviews, and the most advanced models have been barred from public release.