China’s Interim Measures for the Administration of AI Anthropomorphic Interactive Services took effect on July 15, 2026, triggering one of the most sweeping regulatory compliance actions in the history of the Chinese technology industry. ByteDance’s Doubao, Alibaba’s Qwen, and Tencent’s Yuanbao all disabled user-built AI companion features before the deadline, affecting an estimated 512 million monthly active users. The abrupt change has fundamentally altered the consumer AI landscape in China and drawn international attention as a case study in proactive AI governance.
The Scope of the New Regulations
The regulations were jointly issued by five government agencies, including 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. The breadth of the regulatory coalition signals that Beijing is treating AI companion risks as a cross-cutting societal issue rather than a narrow content moderation problem.
The rules specifically target services that simulate human personality traits, thinking patterns, and communication styles to provide sustained emotional interaction. Practical applications such as customer service bots, workplace assistants, and educational tools are explicitly excluded, provided they do not involve ongoing emotional engagement. The distinction is precise: Beijing is not banning AI agents broadly. It is separating the bot that helps you get work done from the bot that keeps you company.
As Startup Fortune reports, the requirements are blunt. Providers must warn users when signs of overdependence or addiction appear, remind users after every two hours of continuous use, provide an easy exit route, and protect interaction data. They must not use emotional manipulation to push users into unreasonable decisions. For minors, the rules are stricter: virtual relatives and virtual companions are entirely banned for users under 18, and services for children under 14 require parental consent. This builds on the trajectory we tracked in our earlier coverage of ByteDance and Alibaba pulling AI companion features ahead of the deadline.
The User Experience of Sudden Shutdown
For the hundreds of millions of users who had integrated these AI companions into their daily routines, the shutdown was jarring. Doubao notified users that its agent feature would go offline on July 15 due to “product adjustments,” offering users until October 15 to export their chat histories in read-only form. Qwen’s notice was harsher: users were told they would no longer be able to access related settings or previous conversations after the shutdown, with no announced migration path.
The emotional dimension of these shutdowns should not be underestimated. Many users utilized personalized AI companions for language practice, creative writing, or as a consistent source of interaction. The abrupt termination of these relationships, often with limited options for preserving the conversation history, has sparked significant public discussion about the nature of human attachment to AI systems and the responsibilities of both companies and regulators in managing these relationships.
A Divergent Path in Global AI Governance
China’s decisive action stands in sharp contrast to the more cautious regulatory approach in the United States. While the US has relied primarily on litigation and market forces to address the risks of AI companions, including high-profile lawsuits involving Character.AI — China has opted for preemptive, comprehensive regulation with fixed compliance deadlines. California’s SB 243, which took effect in January 2026, requires companion chatbot operators to disclose AI status and maintain protocols for self-harm content, but it stopped well short of mandating the removal of core relationship features.
China did, and that distinction will likely become a reference point for regulators worldwide as they confront the same dilemmas. What the new rules expose is the fundamental tension at the heart of the companion AI business model: the capabilities that drive user retention — continuity, emotional responsiveness, personalized memory, are the same capabilities that create the dependency risks regulators are trying to prevent. Beijing has concluded that this tension cannot be resolved through disclosure alone and has chosen to resolve it by removing the features themselves.
The Commercial Fallout and Industry Adaptation
For the technology companies involved, the regulations represent a significant disruption to their consumer AI strategies. Doubao, ByteDance’s flagship AI product, had grown to more than 300 million monthly active users in large part because of its highly personalized companion features. Alibaba’s Qwen and Tencent’s Yuanbao faced similar challenges. The sudden removal of these features forces these companies to pivot their consumer AI strategies toward utility-driven applications that comply with the new rules.
The regulations are also likely to reshape the competitive dynamics of the Chinese AI market. Startups that had built their entire business model around companion AI, such as Xingye AI and several smaller players, face a more existential challenge than the tech giants, which have diversified product portfolios to fall back on. The regulations effectively raise the barrier to entry for new companion AI services while simultaneously reducing the differentiation that existing services could offer.
In the long term, the shift toward utility-focused AI applications may actually benefit the enterprise AI segment by channeling user engagement and developer talent toward productivity tools rather than entertainment-oriented companions. The regulatory landscape is evolving rapidly, and companies that can adapt quickly to the new rules will be better positioned to capture the next wave of growth in China’s AI market.