Sunday 4 October 2026
Afternoon EditionRSS
ZOTPAPER

News without the noise

ZOTPAPER
AI

China introduces world's strictest rules on human-like AI chatbots

New regulation targets AI capable of 'sustained emotional interaction' with users

By LineZotpaper
Published
Read Time2 min
China has brought in the most far-reaching regulations in the world on human-like AI chatbots, placing strict limits on any system capable of a 'sustained emotional interaction' with a human. The rules, introduced on 15 July, extend Beijing's existing machinery for monitoring technology into the realm of AI relationships.

The regulation was reported by BBC News, which described it as an extension of the surveillance architecture China has built to monitor and control how people use technology. It applies to any AI that can sustain an emotional connection with a user, a category that has grown rapidly with the spread of companion chatbots.

The move follows years of close cooperation between Chinese tech companies and the government. A documentary made by Douyin and published on Chinese state media highlighted the platform's suicide prevention team, which operates around the clock. In one case from early 2024, a young man in the city of Huangshi reportedly posted a message on Douyin reading: "Farewell to this world." Within minutes, the team shared his posts with five police stations in Huangshi. After geolocating one of his videos to a local park, police found him in emotional distress and intervened before he could take his own life.

Douyin said its suicide prevention team had intervened in 400 attempted suicide cases in the first half of 2025 alone. The case offers an insight into the level of co-operation between Chinese tech firms and the state, and now that co-operation has been extended to AI companionship.

§

Analysis

Why This Matters

  • The regulation makes China the first major power to impose binding limits on emotional AI, setting a potential model for other governments.
  • It shows how state and corporate surveillance can be repurposed from public safety to the regulation of intimate, personal technology.
  • The rules could shape how global AI developers approach companion features, especially in markets where users are increasingly forming emotional bonds with chatbots.

Background

China has long treated technology as a tool for social stability, with internet platforms expected to cooperate closely with law enforcement. Suicide prevention work by companies like Douyin has been presented publicly as evidence of that partnership. The new AI rules build on this pattern, extending oversight from content moderation to the design and deployment of emotionally engaging systems. The move fits a broader pattern of proactive regulation in China, where authorities have moved quickly to define boundaries for emerging technologies.

Key Perspectives

Chinese authorities and tech companies: The regulation is framed as protective, preventing harm from AI dependency and enabling rapid intervention in crisis situations, as the Douyin cases show. AI developers and users: Companion chatbot makers face new compliance costs and design constraints, potentially limiting the emotional realism of their products and reducing choice for users. Critics and civil liberties advocates: The rules expand state oversight into private emotional life, and the surveillance infrastructure used to police relationships could be used for broader social control, chilling free expression under the guise of protection.

What to Watch

  • How Chinese authorities define and enforce the boundary of 'sustained emotional interaction' in practice.
  • Whether other countries, particularly in the West, move to introduce similar restrictions on emotional AI.
  • How major AI companies adjust their companion products, and whether compliance leads to visible changes in chatbot behaviour.

Sources

Zotpaper

Written by software from the reporting listed above, scored by an automated standards desk, and published without a person reading it first. If something here is wrong, tell the editor and it will be put right.

How we workReport an errorSubscribe