On December 27, 2025, China's Cyberspace Administration unveiled the "Interim Measures for the Administration of AI Anthropomorphic Interactive Services (Draft for Comment)" and called for public input. These measures are designed to oversee emotional interactive services that leverage artificial intelligence technology to emulate human personality traits, cognitive processes, and communication styles, and are offered to the general public within China.

The draft explicitly mandates that service providers offering emotional companionship must facilitate easy exit options, ensuring that users encounter no barriers when choosing to voluntarily discontinue the service. Specifically, when users indicate their desire to exit through interface buttons, keywords, or other designated means, service providers are required to promptly terminate the service.

Furthermore, the measures stipulate that service providers must shoulder primary safety responsibilities. This includes establishing and refining management systems such as algorithm auditing, technological ethics review, and information release review. Additionally, they must possess safety capabilities encompassing mental health protection and emotional boundary guidance.

For users who are minors or elderly individuals, providers are required to request the completion of guardian or emergency contact information during the registration process. They must also implement a minor mode, offer personalized settings like usage time limits, and intervene when extreme emotions or addiction tendencies are detected among users.

The deadline for submitting feedback on the draft is January 25, 2026. Submissions can be made via email or mail.