Chinese authorities are implementing strict controls on generative AI technologies that mimic human emotions and conversational patterns.

On the 27th, the Cyberspace Administration of China released a draft of regulations for AI-based conversational services. This mandate applies to all service providers that simulate human-like interaction through text, voice, or video. Public feedback on the draft is being accepted until January 25th.

The core of the draft focuses on preventing excessive user dependency and addiction. AI service providers must clearly inform users that they are interacting with an AI, not a human, at the time of login and during conversations. In particular, if a user utilizes the service for more than two hours continuously, the provider must trigger a pop-up or warning to encourage the user to stop, and the logout process must be designed to be simple.

Ideological censorship standards for content have also been significantly strengthened. Content that threatens national security or undermines the core values of socialism is strictly prohibited. Furthermore, AI must be trained on datasets that reflect Chinese history and traditional culture without compromising regime stability. Generating information related to violence, gambling, pornography, or defamation is also classified as illegal.

Measures for the physical and mental safety of users have also been included. Content that glorifies self-harm or suicide must be blocked, and systems must be in place to notify guardians or relevant authorities if a user shows signs of extreme distress. Additionally, minors are prohibited from using AI technology without parental consent, and large companies with over 1 million registered users must submit regular assessment reports to the authorities.