China has announced plans to tighten oversight of artificial intelligence systems that act like humans. Companies will have to tell users when they’re interacting with machines and follow strict guidelines on content and security.
The proposed regulations would make providers notify people they’re using AI when they first log in and again every two hours. They’d also need to warn users if the system detects they might be relying on it too much. The Cyberspace Administration of China posted the draft rules on Saturday for public review. Comments are being accepted until Jan. 25.
Companies offering human-like AI must set up strong security checks and ethics reviews. The systems also need to follow what the government calls “core socialist values” and can’t share anything that threatens national security.
China balances AI growth with tight controls
The move shows China’s approach to AI development. The country is pushing hard on the technology to help the economy grow and compete worldwide. As reported by Julisha.co.ke previously, Nvidia boss Jensen Huang said China will leave U.S. behind in AI race.
However, officials also want tight control to prevent disruptions to security or social order.
So what do the new AI governance rules require? Companies must complete a security review and send a report to provincial internet regulators before launching any human-like AI features. They’d also need to file reports if their service reaches 1 million registered users or 100,000 monthly active users.
Many people focus on which country will build the most advanced AI models. But Beijing has already taken the lead in a different area: creating global rules for how AI should work. This effort worries people who care about free speech because Beijing has spent years trying to influence international technology standards and promoting strict internet controls.
China released its Global AI Governance Action Plan on earlier this year on July 26. The detailed strategy tries to make Beijing the main voice in setting international AI rules. It includes proposals like the World Artificial Intelligence Cooperation Organization. Beijing says AI should be a “public good for the international community” managed for “safety” and shared benefits.
Democratic countries have reasons to doubt China’s plans for global AI governance. China already blocks massive amounts of online content through its Great Firewall. Experts say similar controls now apply to AI. Anyone using DeepSeek, a major Chinese AI model, can see how it refuses to discuss topics the government considers sensitive.
China ranks last in free speech protections
Research released in October compared AI policies in six countries. China ranked dead last among major AI nations when it came to protecting free speech in AI. The United States, European Union, Brazil, South Korea, and India all scored better. The findings match what users already know. China’s Communist Party shapes AI in the country to match its political goals and social limits.
Beijing built a complicated network of government rules and company requirements that work together as a governance system. That’s different from the European Union, which passed one main AI law called the AI Act.
The country’s main regulation for AI that creates content is called Interim Measures for the Management of Generative AI Services. It requires companies to support “core socialist values” and block content that might challenge government power, encourage breaking away from China, or disturb economic and social order.
The same regulation says training data must meet strict political standards. While officially calling for “truth,” “accuracy,” and “objectivity,” the rules actually mean companies can’t use data that challenges the current system, damages China’s reputation, creates “harmful” information, or goes against social customs, ethics, or morality.
