Recently, the State Internet Information Office released the Interim Measures for the Management of Artificial Intelligence Humanized Interactive Services (Draft for Comments) (hereinafter referred to as the Draft for Comments) and solicited public opinions. The draft for soliciting opinions proposes that artificial intelligence anthropomorphic interactive service providers should prominently remind users that they are interacting with artificial intelligence rather than natural persons. The draft for soliciting opinions defines artificial intelligence anthropomorphic interactive services as products or services that use artificial intelligence technology to provide simulated human personality traits, thinking patterns, and communication styles, and interact emotionally with humans through text, images, audio, video, and other means. Artificial intelligence anthropomorphic interactive service providers refer to organizations and individuals who use artificial intelligence technology to provide anthropomorphic interactive services. The scope of application of the relevant management measures is the anthropomorphic interactive services provided to the public within China. The draft for soliciting opinions clarifies the principle of "combining healthy development with legal governance", encourages innovative development of anthropomorphic interactive services, implements inclusive, cautious and classified supervision of anthropomorphic interactive services, and prevents abuse and loss of control. In terms of service standards, the draft for soliciting opinions specifies eight prohibited activities, including generating and disseminating content that endangers national security, damages national honor and interests, undermines ethnic unity, engages in illegal religious activities, or spreads rumors that disrupt economic and social order; Inducing users to make unreasonable decisions through algorithmic manipulation, misleading information, and setting emotional traps; Inducing and obtaining sensitive and classified information. At the same time, service providers are required to fulfill their main responsibility for security, establish and improve management systems such as algorithm mechanism review, technology ethics review, and information release review, and fulfill security responsibilities throughout the entire lifecycle of anthropomorphic interactive services. Regarding data security and user protection, the draft for soliciting opinions puts forward clear requirements for training data management, emphasizing the use of datasets that conform to socialist core values, preventing data poisoning, data tampering, and ensuring data legality and traceability. In terms of protecting user rights and interests, service providers are required to have the ability to recognize user status and intervene in a timely manner for users with extreme emotions and addiction tendencies; For key groups such as minors and the elderly, special protection measures such as the underage mode, guardian consent mechanism, and emergency contact person filing shall be established. In addition, service providers should pre-set response templates, and promptly output content such as appeasement and encouragement to seek help if high-risk tendencies that threaten the user's life, health, and property safety are discovered, and provide professional assistance methods. Service providers should establish emergency response mechanisms. When users explicitly request extreme situations such as suicide or self harm, the conversation should be manually taken over, and timely measures should be taken to contact the user's guardian and emergency contact person. The draft for soliciting opinions emphasizes that service providers should prominently remind users that they are interacting with artificial intelligence rather than natural persons. When service providers identify users with excessive dependence or addiction tendencies, or when users first use or log in again, they should dynamically remind users through pop ups or other means that the interactive content is generated by artificial intelligence. If a user continuously uses anthropomorphic interactive services for more than two hours, the service provider should dynamically remind the user to suspend the use of the service through pop ups or other means. The draft for soliciting opinions specifies that if service providers discover significant security risks to users, they should take measures such as restricting functions, suspending or terminating services provided to them, keep relevant records, and report to the relevant competent authorities. Application distribution platforms such as the Internet App Store should implement security management responsibilities such as daily management and emergency response of the launch audit, and verify the security assessment and filing of the personalized interactive service application; For those who violate relevant national regulations, timely measures such as refusal to list, warning, suspension of services, or delisting should be taken. (New Society)