China Plans AI Safety Standards Covering the Full Life Cycle of the Technology

Wang Ruomeng, a senior NDRC official, told a briefing in Beijing that China would accelerate legislation covering how AI is developed and applied. That includes the technology's core inputs of data, computing power and algorithms, as well as key areas where it is used, according to Bloomberg's report, published by The Star. The rules would apply to Chinese developers including DeepSeek, Moonshot and Alibaba Group Holding.

What Beijing Is Proposing

According to the reports, AI developers and operators would have to strengthen risk management and disclose safety risks, incidents and vulnerabilities. To limit the compliance burden, Beijing plans an unspecified transition period and guidance for companies, along with sector-specific regulatory sandboxes that allow some room for trial and error. Officials said the rules would be reviewed regularly and outdated provisions revised.

The NDRC also said the government would closely monitor AI's effect on employment and job displacement, The Information reported, and would provide safety-net protections for affected workers. Beijing has long tried to avoid large-scale social disruption, and that concern now extends to AI as the systems become more capable.

China will keep encouraging open-source AI development through its 15th five-year plan, which runs to 2030, according to Bloomberg. Several of the country's leading labs release their model weights openly.

Wang also called for international cooperation, including deeper mutual trust, coordinated vulnerability disclosure and joint responses to risks. "AI safety is not a problem one company, industry or country can tackle alone," he said, according to Bloomberg.

The Wider Context

The announcement comes amid growing global concern about AI safety, Bloomberg reported, following warnings from Silicon Valley leaders about existential risks and recent incidents in which AI systems behaved in unintended ways or were used to strengthen cyber-breach attempts. In September, China and the United States agreed to try to build a common understanding of AI's benefits and threats, ahead of a summit in Washington between Donald Trump and Xi Jinping at which AI was a key topic.

Outlets describe Wang's role differently. Bloomberg calls him a senior NDRC official, while some other reports name him as a deputy director in the commission's innovation and high-tech department. No draft text of the standards has been published.

Why It Matters

For companies building or deploying AI in China, the plan points to new disclosure and risk-management duties, softened by transition periods and sandboxes. For governments elsewhere, including in the Gulf, where the UAE and Saudi Arabia are writing their own AI governance rules while buying US chips and working with US labs, it shows that Beijing wants a say in global AI safety rules. Its call for coordinated vulnerability disclosure is one area where it could find common ground with Washington.