China plans AI safety rules spanning development and deployment

[Courtesy Bloomberg]

China plans to introduce extensive artificial intelligence safety standards covering the development and deployment of models, as Beijing seeks to strengthen oversight while keeping its technology industry competitive with the United States, Bloomberg reported.

The proposed framework will cover the entire life cycle of AI systems, including data, computing resources and algorithms, Wang Ruomeng, a senior official with the National Development and Reform Commission, said at a briefing in Beijing on Tuesday.

The measures would affect developers including DeepSeek, Moonshot and Alibaba, whose models are central to China’s efforts to compete with American companies such as OpenAI and Anthropic.

According to Bloomberg, the initiative reflects Beijing’s attempt to balance technological ambition with concerns that increasingly capable systems could escape human control or make cyberattacks more damaging. Warnings from technology industry leaders about existential risks have added to international pressure for stronger safeguards.

Wang said authorities would accelerate legislation governing the foundations of AI and its use in important sectors, while introducing ethical guidelines and safety standards. Officials would regularly assess implementation and revise rules that no longer reflect technological developments.

Beijing also plans measures to help companies adapt. These include a transition period, whose duration was not specified, and guidance intended to reduce the burden of compliance.

Sector-specific regulatory sandboxes would give developers room to test applications and learn from failures under supervision. The approach signals an effort to preserve experimentation alongside tighter oversight, rather than impose uniform restrictions across every use of AI.

Developers and operators, however, would be expected to strengthen their own risk management and disclose safety threats, incidents and vulnerabilities, Wang said.

The proposals also address employment. Authorities will monitor how AI changes the labor market and displaces workers, and provide broad protections for those affected. Wang’s comments underline Beijing’s concern that rapid adoption could cause social disruption as automated systems become capable of performing more tasks.

Despite the planned safeguards, China will continue encouraging open AI development during its 15th five-year plan, which runs through 2030. That commitment preserves an approach that has helped Chinese models gain users internationally, Bloomberg reported.

The announcement comes against a backdrop of discussions between Washington and Beijing about managing the technology’s risks. China and the United States agreed in September to pursue a shared understanding of AI’s benefits and threats, ahead of a Washington summit between Presidents Donald Trump and Xi Jinping at which AI featured prominently.

Wang urged governments to strengthen cooperation on safety governance, coordinate the disclosure of vulnerabilities and responses to risks, and build greater confidence in one another’s systems.

His remarks presented international coordination as essential to managing a technology whose development and effects extend beyond national borders. They also paired China’s domestic regulatory agenda with an appeal for cooperation despite competition between the world’s leading AI powers, which are both pursuing increasingly powerful artificial intelligence systems.

“AI safety is not a problem one company, industry or country can tackle alone — it is a global challenge requiring international cooperation,” Wang said, according to Bloomberg.