China: Cyberspace Administration closes consultation on draft Interim Measures for the Administration of Anthropomorphic Interactive Artificial Intelligence Services including data protection regulation

On 25 January 2026, the Cyberspace Administration of China closes the consultation on the draft Interim Measures for the Administration of Anthropomorphic Interactive Artificial Intelligence Services, including data protection regulation. The measures apply to anthropomorphic interactive AI services, defined as services that simulate human personality, thinking patterns, and communication styles and interact with humans through text, pictures, audio, video, or by other means. Article 10 would establish data protection requirements, including using training data conforming to core socialist values, cleaning and labelling training data to enhance transparency and reliability, ensuring the diversity of training data through negative sampling and adversarial training, evaluating the security of synthetic data, regularly upgrading data, and ensuring that training data is legal and traceable. Further, providers of anthropomorphic interactive services would need to take appropriate data security measures, refrain from providing interaction data to third parties, allow users to delete interactive data, and obtain additional guardian consent from users who are minors.