The Cyberspace Administration of China, along with four other departments, has jointly released the Interim Measures for the Administration of Artificial Intelligence Human-like Interaction Services. These measures focus on regulating the following key areas:
Firstly, clarifying primary safety responsibilities.
Secondly, strengthening safety management throughout the entire lifecycle of human-like interaction services. This entails setting clear safety requirements at each stage—deployment, operation, upgrading, and termination—to ensure that safety measures are implemented concurrently with service functions, thereby enhancing overall safety levels. Simultaneously, it involves reinforcing safety monitoring and risk assessment to promptly rectify system deviations and address safety incidents as they arise.
Thirdly, safeguarding data security. This is achieved by enhancing the management of training data, improving data transparency, reliability, diversity, and security. It also involves implementing data property rights systems in accordance with the law and adopting measures such as data encryption and access control to protect user interaction data. Without legal authorization or consent from the right holder, user interaction data must not be provided to third parties.
Fourthly, protecting user rights and interests. The measures stipulate that providers of human-like interaction services must fulfill their obligations to protect users' personal information and privacy. They must offer options for copying and deleting interaction data, as well as convenient avenues for service exit. When discontinuing services, users must be notified in advance or announcements must be made promptly. Additionally, a robust mechanism for handling user appeals and public complaints and reports must be established.