The Cyberspace Administration of China (CAC), together with four other government authorities, has issued the Interim Measures for the Administration of Anthropomorphic AI Interactive Services (Joint Order No. 21), which entered into force on 15 July 2026, establishing a regulatory framework for AI services providing continuous emotional interaction through the simulation of human characteristics, thought patterns and communication styles.
The Measures apply to services offering emotional care, companionship or support through text, images, audio or video, while excluding applications that do not involve continuous emotional interaction, such as conventional customer service, knowledge-based assistance and educational tools. They impose obligations concerning content governance, personal information and interaction data protection, the prevention of excessive emotional dependence, and safeguards for minors and elderly users. Providers must also implement appropriate safety measures, clearly inform users that they are interacting with AI rather than a human being, and conduct security assessments under the circumstances specified by the Measures.
For developers and providers of services falling within their scope, the Measures establish specific compliance requirements covering service design, user protection, data management and ongoing risk assessment. The framework places particular emphasis on managing the risks associated with sustained emotional interaction between users and AI systems, while distinguishing such services from other forms of artificial intelligence assistance.