China is rushing through a new law that would put artificial intelligence (AI) under government supervision at every stage, from the data and computers used to build it to the way it is used.
The plan was announced by the National Development and Reform Commission (NDRC), the country's main economic planning agency.
Wang Ruomeng, a senior official at the agency, said the rules would be reviewed regularly and updated as the technology changes.
The law is expected to cover China's biggest AI developers, including DeepSeek, Moonshot AI and Alibaba.
All three are building chatbots and other AI products designed to compete with those from OpenAI and Anthropic in the US.
Safety and testing
Companies that build or run AI systems would have to do more to manage risks.
They would also have to report safety problems, incidents and weaknesses in their systems.
At the same time, regulators plan to create testing zones for different industries, where firms can try out AI before permanent rules are set.
Protecting workers
The law will also look at the impact of AI on jobs.
Beijing plans to track how many workers lose their jobs to AI and to provide support for those affected.
A rapid build-out
The move comes as China expands its AI infrastructure at speed.
Its AI computing capacity reached 2,185 EFLOPS by the end of June, a measure of how many calculations its computers can perform each second.
That was up 177% on a year earlier, and a national plan aims to reach 9,800 EFLOPS by 2030.
Read more about AI Regulation
- 19/09/26China says we are looking at AI safety all wrong. Silicon Valley should pay attention
- 14/09/26China dismisses AI slowdown calls as "fearmongering" aimed at containing Beijing
- 11/05/26China's agentic AI regulations are the most detailed in the world, and the West has nothing comparable
China had also built more than 120,000 high-quality datasets by mid-2026, the collections of information used to train AI.
More than 700 million people in China now use AI tools that can generate text, images and other content.
With so many users, risks such as misuse, privacy breaches and job losses have become a national issue.
Export curbs and US talks
The plan follows draft rules in October that reportedly limit exports of 23 types of advanced AI technology, including chatbot models, AI agents and large computer systems.
It also comes two months after President Donald Trump and President Xi Jinping met in Washington in September.
The two leaders agreed to share views on the risks of AI and set up a hotline for dealing with AI-related accidents.
