Draft interim measures would govern AI services that mimic human traits and interactions.
The Cyberspace Administration of China (CAC) on December 27 issued a draft interim measure to regulate “human-like artificial intelligence interactive services” and opened it for public comment through January 25, 2026. The notification and draft text were published on the CAC’s official website.
The proposed Interim Measures for the Administration of Human-Like AI Interactive Services would apply to products and services that use artificial intelligence to simulate human personality traits, thinking patterns, and communication styles to interact emotionally with people via text, images, audio, or video within China.
Under the draft, AI service providers must comply with specified service standards and safety obligations, including mechanisms for algorithm review, ethical assessment, data handling, cybersecurity, personal information protection, anti-fraud measures, and risk response planning. The measures would require providers to clearly indicate when users are interacting with AI rather than a human and to implement protections to prevent harmful content, emotional manipulation, and over-dependence.
The draft also includes provisions to safeguard minors and older users, such as age-specific modes, usage duration reminders, consent requirements for emotional companion services, and mechanisms to monitor and intervene in cases of excessive use or high-risk user behavior.
According to the CAC notice, the draft is based on China’s Civil Code, Cybersecurity Law, Data Security Law, Personal Information Protection Law, and other legislation. It proposes inclusive, prudent, and tiered supervision designed to balance innovation with legal compliance and public safety.
An official CAC statement accompanying the draft said the measures aim to “promote healthy development and standardize application” of human-like AI services. “Providers should build secure and controllable technical safeguards … and respect social ethics,” the statement said.
