China to Tighten Regulations on Human-Like AI

China’s cybersecurity regulator has released a draft of rules for public consultation, aimed at overseeing AI services that mimic human personalities, according to Reuters.
The regulations will apply to AI-based products and services that replicate human-like thinking models and communication styles, interacting with users on an emotional level. Communications may involve texts, images, audio, video, and other materials.
The document is intended to strengthen safety and ethical requirements for providers of such services.
The draft rules require service providers to:

• be responsible for safety throughout the product’s lifecycle;
• implement an algorithm verification system to protect data and private information;
• assess users’ emotional states, their level of psychological dependency on the service, and intervene when necessary.

The proposed measures also prohibit AI services from generating content that threatens China’s national security, spreads rumors, or promotes violence and obscenity.
In 2021, China’s Ministry of Science and Technology introduced a set of ethical principles for AI regulation, focusing on protecting user rights and preventing risks. Key points included ensuring AI’s “manageability and reliability,” fairness, and enhancing ethical literacy.
Two years later, the Cyberspace Administration published regulations for generative AI algorithms. Requirements included ensuring content accuracy and respecting intellectual property, with a specific mention of prohibiting calls to undermine the state system.
Earlier, Chinese authorities perceived a threat of overheating in the humanoid robot market.