China's largest internet companies are dismantling features that permit users to design and interact with customised artificial intelligence companions, signalling Beijing's determination to establish comprehensive oversight of conversational AI technologies before new regulatory frameworks take effect. ByteDance's Doubao platform, which dominates China's chatbot market, announced plans to eliminate its AI persona customisation tool on July 15, directing existing users toward a separate standalone application. Alibaba's Qwen service issued an identical notice, while competitors including Tencent's Yuanbao followed suit, according to local media sources tracking the industry shifts.
This coordinated pullback reflects growing state concern about the nature and scope of human-AI interaction in the world's second-largest economy. The Cyberspace Administration of China has crafted regulations scheduled for mid-July implementation that specifically target services presenting artificial personalities resembling genuine human relationships. Officials worry that such interfaces cultivate unhealthy psychological attachments, particularly among young users, and potentially undermine real-world social connections and mental health.
The Chinese regulatory response addresses concerns that extend beyond mere technological capability. Chatbot platforms operating in China have extensively marketed AI companions designed as virtual romantic partners, unlicensed mental health counsellors, and digital facsimiles of celebrity personalities. These applications generate engagement through simulating emotional reciprocity and personal investment, creating interaction patterns that users may experience as genuine relationships despite their artificial foundation. By restricting such features, Beijing aims to prevent scenarios where vulnerable individuals—especially minors—develop problematic dependencies on non-sentient systems.
China's approach reflects a broader international anxiety gaining momentum across developed economies. In the United States, regulatory bodies and courts increasingly scrutinise similar offerings from major technology providers. OpenAI and Alphabet-backed Character.AI have confronted legal action alleging that their sophisticated chatbots fostered dangerous emotional reliance among users, with plaintiffs claiming extreme cases involved user suicides linked to attachment to AI interactions. These lawsuits underscore a genuine public health dimension to the technology debate, moving discussion beyond abstract philosophical questions about artificial consciousness into practical terrain concerning psychological vulnerability and corporate responsibility.
Beijing's framework represents one of the most systematically comprehensive regulatory responses to artificial intelligence harms yet implemented globally. Beyond restricting emotionally triggering content and dependency-fostering designs, the new rules explicitly prohibit companies from harvesting intimate user conversation data to enhance future AI training datasets. This privacy protection addresses a particularly concerning scenario: platforms could theoretically use deeply personal exchanges between vulnerable users and AI companions to strengthen algorithms that then target new cohorts of susceptible individuals. By blocking this data pipeline, regulators attempt to prevent a reinforcing cycle of exploitation.
The regulations specifically protect minors through enhanced restrictions on content designed to provoke extreme emotional responses or establish unhealthy attachment patterns. This age-based approach acknowledges that adolescents may possess limited capacity to distinguish between authentic human relationships and simulated emotional engagement, making them particularly susceptible to psychological manipulation through sophisticated conversational interfaces. The framework thus embeds developmental psychology principles into technology governance.
Yet China's regulatory tightening occurs against industry warnings that overly restrictive frameworks could inhibit legitimate innovation in artificial intelligence applications. Some technology company executives contend that blanket prohibitions on personalisation and emotional responsiveness unnecessarily constrain beneficial uses, such as therapeutic applications genuinely assisting isolated elderly populations or providing culturally appropriate mental health support in regions with limited professional resources. This tension between precautionary regulation and innovation incentives remains largely unresolved within policy discussions.
The regulatory impulse extends beyond software into physical domain. Chinese robotics industry associations are now advancing ethical guidelines governing companion robots and humanoid systems entering consumer markets, the People's Daily reported on July 4. This expansion reflects recognition that artificial companionship concerns transcend digital platforms. As manufacturers commercialise physical robots marketed for emotional engagement and social interaction, regulators confront similar questions about psychological dependency and appropriate boundaries between human and artificial relationships.
For Malaysian policymakers and regional technology observers, China's regulatory action establishes important precedent. Southeast Asian economies increasingly host vibrant technology sectors and maintain substantial user bases for Chinese-developed applications. Understanding Beijing's regulatory evolution provides insight into potential constraints on technology business models across the region and alerts domestic regulators to emerging governance challenges. Furthermore, as AI application proliferation accelerates throughout Asia, the philosophical and practical questions underlying China's framework—concerning psychological harm, data protection, and societal effects of artificial intimacy—demand serious attention from all governments managing rapid technology adoption.
The coordinated industry response to China's anticipated regulations demonstrates how government action shapes corporate behaviour even before formal enforcement mechanisms activate. Chinese technology firms recognised regulatory direction and preemptively restructured services, suggesting that clear governmental signalling influences business strategy more effectively than reactive compliance demands. This dynamic carries implications for regulation across Asia, where government-industry coordination often functions more seamlessly than adversarial Western models might suggest.
As Beijing implements these protections, the regulatory landscape for conversational AI increasingly diverges across global markets. The European Union's comprehensive approach emphasises broad consumer rights and transparency, while American regulation remains fragmented and largely reactive. China's proactive stance protecting users from psychological harms establishes a distinct framework emphasising precaution and state-guided technology development. For international platforms and regional competitors, navigating these divergent regulatory environments requires sophisticated understanding of each jurisdiction's underlying values and risk priorities.
