The immediate consequence for users of ByteDance's Doubao and Alibaba's Qwen is the loss of their custom AI agents and any associated data unless it is exported before July 15, 2026. This date marks the full implementation of China's new regulations on "anthropomorphic AI interaction services." For the broader AI industry in China, the shutdown indicates a clearer, more restrictive regulatory environment for AI applications that foster sustained emotional interaction or simulate human-like personalities. Companies will likely pivot their development efforts towards enterprise and productivity-focused AI to avoid similar compliance issues.
Image: courtesy of Thenextweb
China's AI Companion Crackdown: What Beijing's New Rules Mean for User Creativity and the Tech Giants
ByteDance's Doubao and Alibaba's Qwen are disabling custom AI companion features by July 15, 2026, in response to new Chinese regulations targeting "anthropomorphic AI interaction services." Users face permanent data loss if they do not export their custom agents before the deadline. This move signals a significant tightening of control over emotionally-oriented AI, drawing a clear line between general assistants and personalized AI companions within China's digital ecosystem.
Outlook
Background
The decision by ByteDance and Alibaba to pull user-created AI agents comes directly from China's new, world-first regulations specifically targeting "humanlike AI interaction." While the precise text of these rules has been subject to various interpretations, the core focus appears to be on AI services that allow for or encourage sustained emotional engagement, simulate personalities, or retain relationship context over time. This distinguishes them from more general AI assistants or productivity tools.
The regulatory push does not exist in a vacuum. It sits alongside China's already robust framework of digital control, which includes the Cybersecurity Law (2017), the Data Security Law (2021), and the Personal Information Protection Law (2021). These laws grant Chinese authorities broad access to data held by domestic companies and impose strict controls on cross-border data transfers. The National Intelligence Law (2017) further obligates companies to cooperate with intelligence work. The new AI companion rules appear to extend this existing control into the realm of artificial intelligence, particularly where AI could influence personal sentiments or generate user-created content that might be deemed sensitive.
See also
Precedents
China has a well-established history of regulating its internet and technology sectors, often starting with broad directives and then refining them as new technologies emerge. This pattern has been evident in areas from social media content to online gaming and data privacy. Historically, the emphasis has been on maintaining social stability, controlling information flow, and ensuring data security within the state's purview. The regulatory response to AI companions fits this mold, extending control to new forms of user-generated content and interaction.
Past regulatory cycles have shown that Beijing frequently prioritizes control over unfettered innovation, especially in areas touching on public sentiment or the dissemination of information. While experts suggest regulators aim to balance control with promoting Chinese AI development, the initial approach often leans towards caution. This typically leads to a period of adjustment for tech companies, where features are removed or modified before a more stable, compliant operational model emerges. The current crackdown on AI companions is a direct continuation of this long-standing approach to digital governance.
This isn't just a technical adjustment for two major Chinese tech firms; it signals a fundamental shift in how AI can be developed and used in one of the world's largest digital markets. The removal of custom AI companion features highlights Beijing's proactive stance on AI governance, particularly concerning the psychological and social implications of human-AI interaction.
For users, it represents a loss of creative freedom and personalized digital experiences. The requirement to export data before deletion also raises questions about the long-term viability and ownership of user-generated content within AI platforms. For ByteDance and Alibaba, it means re-evaluating their AI strategies, potentially pushing them further into enterprise-grade AI or more narrowly defined consumer applications that steer clear of emotional interaction.
More broadly, this move deepens the divergence between China's AI development path and that of Western nations. While some Western regulators are beginning to consider the ethical implications of AI companions, China has moved swiftly to impose strict, pre-emptive controls. This could lead to different types of AI products and user experiences emerging in various global markets, with China's ecosystem potentially favoring more controlled, less personalized, and less emotionally engaging AI applications.
Scenarios
AnalysisOne immediate outcome is a chilling effect on the development of emotionally nuanced AI companions within China. Companies may now shy away from investing in or launching AI products that could be perceived as fostering deep emotional bonds, opting instead for applications with clear productivity or utility functions to avoid regulatory scrutiny. This could stifle innovation in a specific, yet growing, segment of the AI market.
A second outcome could be a refined regulatory framework over time. As regulators observe market reactions and the practical implications of these initial rules, there is a possibility they might issue more detailed guidelines or clarifications. This could offer companies a clearer path for developing certain types of personalized AI, provided they adhere to strict content and interaction protocols, though this is speculative and depends on the regulators' ongoing assessment.
A third, more macro outcome is the deepening divergence of global AI development. China's assertive regulatory approach could lead its domestic AI industry to prioritize different research and product development areas compared to regions like the United States or Europe. This might result in a unique Chinese AI ecosystem that excels in specific areas like industrial automation or large-scale data processing, while personalized, emotionally intelligent AI agents become less prevalent or are developed under very different constraints.
Timeline
2017
China's Cybersecurity Law and National Intelligence Law Enacted
These foundational laws established broad governmental powers over data and mandated cooperation from companies on intelligence matters.
2021
Data Security Law and Personal Information Protection Law Enacted
Further tightened regulations on data handling, storage, and cross-border transfers, impacting how tech companies manage user information.
Early 2026
New Regulations on Anthropomorphic AI Interaction Services Finalized
Beijing introduced new, specific rules targeting AI applications capable of sustained emotional interaction or simulating human-like personalities.
On or Before July 5, 2026
ByteDance and Alibaba Announce AI Companion Shutdowns
ByteDance's Doubao and Alibaba's Qwen announced they would disable custom AI agent features to comply with the impending regulations.
July 15, 2026
Deadline for User Data Export and Regulation Implementation
Users must export their custom AI agent data before this date to avoid permanent loss. The new anthropomorphic AI interaction service regulations take full effect.
Frequently Asked Questions
These are AI services designed to simulate human-like personalities, engage in sustained emotional interaction, or maintain relationship context with users. The new Chinese regulations appear to specifically target these types of AI applications, distinguishing them from more general-purpose AI assistants.
Discussion
Be the first to share your thoughts.