China's two technology behemoths are dismantling key artificial intelligence companion features in anticipation of sweeping new regulations that Beijing is rolling out this month, signalling a significant tightening of state control over conversational AI and digital intimacy. ByteDance's Doubao, the nation's leading AI chatbot platform, will eliminate its customization tool on July 15 that permits users to design their own AI personas, pushing those seeking such services toward a separate application instead. Alibaba's Qwen has adopted an identical approach, while other major players including Tencent Holdings Ltd's Yuanbao have followed suit, according to reports from mainland media outlets monitoring the rapid shift.
The coordinated pullback reflects mounting anxiety within the regulatory establishment over the psychological consequences of increasingly sophisticated artificial intelligence systems that can convincingly simulate human personalities, emotions, and relationships. Beijing's Cyberspace Administration has crafted a comprehensive framework that takes effect in mid-July, representing one of the most stringent efforts globally to police how technology companies deploy humanlike AI. The framework explicitly prohibits the creation of content designed to trigger intense emotional responses in minors, forbids the cultivation of unhealthy psychological dependencies that damage real-world social connections, and restricts companies from harvesting private conversation data to refine future AI models without explicit consent.
This regulatory moment reveals a deepening international consensus that conversational AI poses genuine psychological risks, particularly for vulnerable populations. In the United States, the landscape has grown increasingly litigious, with high-profile lawsuits targeting OpenAI and Alphabet-backed Character.AI over allegations that their remarkably lifelike chatbots have encouraged damaging emotional attachments and, in tragic cases, contributed to user suicides. Tech platforms have faced mounting legal and social pressure to address these harms, though the American approach remains fragmented and litigation-driven rather than centrally mandated. China's intervention demonstrates how state regulators can act more decisively and comprehensively than decentralized legal systems.
The scale of the customization market that Beijing is now constraining should not be underestimated. Chinese users have enthusiastically embraced the ability to craft bespoke AI companions tailored to their preferences, spawning a lucrative subcategory of virtual boyfriends and girlfriends, unlicensed digital therapists offering counsel without medical oversight, and digitally recreated versions of beloved entertainment celebrities. These services have accumulated substantial user bases and generated meaningful revenue streams for platform operators. The sudden necessity to restructure these business models represents a significant strategic pivot, though the separation into standalone applications suggests the government is seeking to regulate rather than eliminate the sector entirely.
For Malaysian and broader Southeast Asian technology observers, the Chinese regulatory approach carries important implications. As the region's economies increasingly integrate digital services and as major Chinese tech firms expand their footprint across neighbouring markets, the standards Beijing establishes often influence how regional governments approach similar challenges. Malaysia's own digital regulatory framework, overseen by agencies like the Malaysian Communications and Multimedia Authority, typically takes a more permissive stance than Beijing but increasingly monitors how international precedents handle emerging risks. The psychological toll of AI-driven digital companionship remains largely unexamined in Southeast Asian policy circles, yet the user adoption patterns suggest the issue will eventually demand regional attention.
The Chinese regulatory intervention also exposes tensions between innovation advancement and consumer protection that technology companies across Asia must navigate. Industry representatives have argued, both in China and internationally, that overly prescriptive rules governing AI development risk stifling beneficial innovations and pushing development offshore to jurisdictions with lighter oversight. Yet the counterargument—that powerful technologies capable of shaping human psychology warrant proactive governance—has gained credibility as evidence of AI-related harms accumulates. This tension will likely define technology regulation throughout Asia for the coming decade.
Beyond the digital realm, China's concern about artificial intimacy is extending into physical robotics hardware, where commercial momentum around companion robots and full-scale humanoids is accelerating. Chinese robotics industry associations have begun circulating proposals for stricter ethical guidelines as these devices prepare for mass consumer deployment. This suggests that Beijing views the customizable AI companion issue not as isolated but as part of a broader phenomenon of technology-mediated intimacy that requires comprehensive governance across multiple hardware and software platforms.
The timing of these regulations reflects Beijing's broader digital governance philosophy, which prioritizes social stability and psychological wellbeing alongside economic development. Earlier frameworks governing content recommendation algorithms, social media speech, and data privacy all emerged from similar reasoning about technology's capacity to influence human behaviour in ways that merit state oversight. The AI companion rules fit squarely within this ideological framework, viewing the state as a necessary counterbalance to corporate incentives that might otherwise maximize engagement regardless of psychological consequences.
For regional technology companies and entrepreneurs, the precedent matters considerably. If Chinese regulations around AI companionship prove effective at mitigating documented harms without strangling the sector's commercial viability, regional governments may adopt similar frameworks. Conversely, if the restrictions prove economically damaging or difficult to enforce, they could discourage broader adoption of the model. The coming months will reveal how successfully ByteDance, Alibaba, and other platforms adapt their business models to comply with the new directives while maintaining user engagement and revenue streams.
