Starting July 15, 2025, temporary measures regulating artificial intelligence-based services that simulate human personality and support sustained emotional interaction come into effect in China. This move has already led giants like ByteDance and Alibaba to begin disabling custom AI agent features in their flagship products — Doubao and Qwen, respectively.
The document, adopted on April 10 jointly by the Cyberspace Administration of China, the National Development and Reform Commission, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation, introduces strict rules for services that use AI to reproduce personality traits, thinking styles, and communication manners of a person, providing emotional support through text, images, audio, or video. Importantly, the regulations do not affect standard support systems, Q&A systems, work assistants, or educational and scientific tools, provided they do not involve sustained emotional interaction.
Key Requirements of the New Rules
Providers of such services must clearly warn the user that they are interacting with AI, not a human. If the system detects signs of excessive dependence or addiction, it must display additional notifications. For continuous use exceeding two hours, the service must remind the user of the session duration. The regulations also require a convenient exit mechanism: if the user requests to end the interaction via interface, voice command, or key phrase, the service must immediately stop and not attempt to retain the person.
Special attention is given to crisis situations. If the tool detects strong emotional reactions, signs of self-harm, suicidal intentions, threats to life and health, or significant financial losses in a user, the provider must take intervention measures, including assisting the user and contacting a guardian or emergency contact. Minors are strictly prohibited from being provided with forms of virtual close relationships, including relatives or partners. For users under 14, any other types of "humanized" AI interaction require parental or guardian consent.
Security and Responsibility
Companies are required to undergo a security assessment when launching a new "humanized" AI service, adding such features, or making significant technological changes. The same obligation arises if the service reaches 1 million registered users or 100,000 active monthly customers. The security assessment report must be submitted to the provincial cyberspace regulatory authority. Authorities may also require an assessment if they see risks to national security, public interests, or user protection.
Violations may result in warnings, requirements to correct the service's operation, suspension of user registration, or blocking of individual functions. Fines of up to 100,000 yuan (approximately $14,730) and up to 200,000 yuan ($29,459) if harm to life and health occurs are also possible.
Market Reaction: Doubao and Qwen Disable Features
The AI assistant Doubao will disable the AI agent feature starting July 15 due to a "product adjustment." After October 15, related data will be processed in accordance with ByteDance's privacy policy and will become unavailable for viewing or recovery in the app. Qwen from Alibaba Cloud will disable "humanized interactive agents" and user-created agent features on July 10, and will cease operations of broader agent services on July 15. Customers will lose access to settings and previous conversations. Earlier, Tencent removed a similar feature from Yuanbao. These decisions sparked heated discussions on Weibo, where some users reacted negatively to the loss of long-term communication with agents.
Enforcement Practice: Crackdown in Shanghai
Preparation for the new rules began early. On June 26, the Shanghai Cyberspace Administration reported the first phase of the "Qinglang" campaign against violations in AI applications. Since late April, the regulator and 17 platforms have removed or blocked over 4.87 million units of illegal and harmful content, processed over 18,000 violating accounts, and removed over 14,000 non-compliant AI agents from platforms. The agency provided clarifications to nearly 100 key platforms, requiring them to strengthen age verification for minors, prevent emotional dependence, and consumer pressure.
As an expert, I see this as a logical but extremely harsh step. China is effectively acknowledging that emotional attachment to AI agents is not just a technological trend but a potential social threat. These measures will set a precedent for global regulation, and other countries, including the US and EU, will have to consider similar restrictions. The AI companion market, valued in the billions of dollars, will now face serious barriers, which may slow its growth but could potentially make it safer.