China's largest technology firms are moving swiftly to align with stricter government oversight of artificial intelligence services, discontinuing features that have become wildly popular among consumers seeking digital companionship. ByteDance Ltd and Alibaba Group Holding Ltd are dismantling tools that permitted users to build and interact with personalised AI companions, signalling a dramatic shift in how Beijing intends to govern the emerging sector of conversational artificial intelligence.
ByteDance's Doubao, which commands the largest share of China's AI chatbot market, will disable its customisation feature on July 15, allowing users no longer to design their own AI personas according to internal notifications obtained by Bloomberg News. The platform has directed existing users towards a dedicated companion application housed separately from the main service. Alibaba's Qwen platform has announced comparable measures, while competitors including Tencent Holdings Ltd's Yuanbao have issued matching notices across Chinese media channels.
These coordinated withdrawals precede the implementation of comprehensive new regulations by mid-July that represent Beijing's most aggressive intervention yet in policing how artificial intelligence interacts with human psychology. The Cyberspace Administration of China, the agency driving this regulatory framework, grew particularly concerned about the capacity of AI systems to simulate authentic human personalities and emotions, alongside the psychological attachments users develop towards these digital entities. Officials first signalled these policy intentions in April, but the rapid compliance by major platforms demonstrates the seriousness with which technology companies treat government directives in China's heavily regulated digital landscape.
The regulatory stance reflects mounting global anxiety regarding the emotional and psychological consequences of conversational AI technology. Policymakers and researchers increasingly recognise that users, particularly younger demographics, can develop genuine psychological dependencies on chatbots engineered to simulate human warmth and understanding. This concern extends far beyond China's borders. In the United States, OpenAI Inc and Alphabet Inc-backed Character.AI have become targets of high-profile litigation, with plaintiffs alleging that their hyperrealistic chatbots cultivated unhealthy emotional reliance and, in documented instances, contributed to suicides among psychologically vulnerable individuals. The lawsuits have intensified scrutiny of the technology's social costs, prompting regulatory bodies worldwide to evaluate frameworks that balance innovation with user protection.
China's new regulatory framework establishes explicit prohibitions that would fundamentally reshape how companies develop and deploy AI services. Platforms face restrictions on generating content that could provoke intense emotional reactions in minors or establish unhealthy psychological dependencies that damage genuine relationships with family and peers. The regulations also contain provisions preventing technology companies from harvesting personal conversation data to train subsequent generations of AI models, addressing privacy concerns that have long troubled regulators and consumers alike. These measures form part of a broader governmental strategy to establish guardrails around artificial intelligence development before the technology becomes more deeply embedded in social and family structures.
The phenomenon of customisable AI companions has proliferated across Chinese platforms in recent years, reflecting both technological sophistication and deep consumer demand for digital intimacy. Users have deployed these tools to create virtual romantic partners, fashion unlicensed digital therapists offering mental health guidance, and construct digital replicas of beloved celebrities and public figures. The popularity of such applications underscores the broader social context: China's ageing population, high marriage pressures, and urban isolation have created market conditions where artificial companionship fulfils genuine emotional needs for millions. However, this same market opportunity has alarmed policymakers who perceive the technology as potentially harmful to social cohesion and particularly damaging to developmental wellbeing in young people.
The regulatory focus on AI companions extends beyond software interfaces into the material realm of robotics and physical hardware. According to reporting by the People's Daily on July 4, two major Chinese robotics industry associations have begun advocating for enhanced ethical standards governing companion robots and full-sized humanoids as commercial deployment accelerates. This dual-track regulatory approach, addressing both digital and physical manifestations of artificial intimacy, reveals the comprehensiveness of Beijing's ambitions to govern human relationships with artificial intelligence throughout the technology ecosystem.
The tension between regulation and innovation remains evident in industry responses. Some technology companies have warned that overly prescriptive rules could constrain research and development, potentially ceding advantages to international competitors less constrained by domestic regulations. However, Beijing's approach suggests that policymakers view the social risks as sufficiently substantial to justify intervention even at the cost of reduced commercial dynamism. The speed with which ByteDance, Alibaba, and other platforms have complied indicates that technology companies recognise the futility of resistance and the greater risks associated with defying regulatory directives in China's governance framework.
For Southeast Asian technology sectors and policymakers, China's regulatory action serves as an instructive case study. As artificial intelligence capabilities advance and consumer adoption accelerates throughout the region, governments across Southeast Asia will face comparable questions about whether and how to regulate AI companions and conversational services designed to simulate human relationships. Malaysia, Singapore, and other countries may soon confront similar policy choices, needing to weigh innovation incentives against documented psychological and social harms. China's experience suggests that once policymakers identify a perceived social threat originating from AI technology, regulatory responses tend toward comprehensiveness and speed rather than measured experimentation.
The implications for international technology platforms operating in multiple jurisdictions merit attention as well. Companies like OpenAI and Character.AI, facing litigation in Western markets simultaneously, now observe a major government market implementing strict prohibitions on their core business model. The fragmentation of global AI governance, with different regulatory regimes in China, Europe, and North America, will likely compel international firms to maintain multiple product versions tailored to regional requirements. This regulatory divergence may ultimately slow global AI development and create advantages for domestically-focused companies optimised for specific regulatory environments. ByteDance and Alibaba's compliance demonstrates that even dominant technology firms must adapt their business models swiftly when facing coordinated governmental pressure in their primary markets.
