China's New Anthropomorphic AI Rules Force ByteDance, Alibaba, and Tencent to Kill Companion Chatbot Features
China's Interim Measures for AI companion services took effect July 15, prompting ByteDance, Alibaba, and Tencent to shut down persona chatbots and leaving users mourning lost digital partners.
Overview
China’s first nationwide rule targeting AI companion chatbots took effect July 15, and it has already forced the country’s largest tech platforms to shut down the persona-based features that let millions of users build ongoing relationships with virtual partners, according to the Associated Press. ByteDance, Alibaba, and Tencent all pulled the AI companion services from their flagship apps to comply, leaving some users mourning the loss of digital relationships that had run for years.
What We Know
The rule, formally China’s Interim Measures for the Administration of Anthropomorphic AI Interaction Services, was issued April 10, 2026, by the Cyberspace Administration of China and four partner agencies, including the Ministry of Industry and Information Technology, and took effect July 15, according to IAPP. The measure defines companion services as technologies that “simulate the personality traits, thinking patterns, and communication styles of natural persons” while providing “emotional care, companionship, support, and other interactive services provided through text, images, audio, video, and other forms,” according to IAPP. Work assistants and education or research tools are exempted.
According to the Associated Press, the regulation bans AI platforms from generating content that could lead users to make questionable decisions through manipulation of their feelings, or that could trigger extreme emotions or unhealthy habits in children and teens. Platforms must also deliver risk warnings against relying excessively on the apps; the rule specifies that “replacing social interaction,” or serving as a substitute for talking to people in person or by phone, cannot be one of the objectives of an online service, the AP reported. Separately, Just Security reports that Article 8 of the rule bars providers from “excessively catering to users” in ways that induce emotional dependence or addiction, or that damage users’ real relationships. Providers are also barred from offering minors virtual intimate relationships such as virtual partners or relatives, and anyone providing other anthropomorphic services to a user under 14 needs parental consent, per Just Security.
The Cyberspace Administration of China oversees compliance, with fines of up to 100,000 yuan for simple violations and up to 200,000 yuan for violations where “harm to citizens’ life and health and safety is accompanied by harmful consequences,” according to IAPP.
ByteDance’s Doubao and Alibaba’s Qwen — described by Just Security as China’s two largest consumer AI apps, with roughly 345 million and 166 million monthly active users respectively as of March 2026 — announced in early July that their AI agent creation features would be discontinued on July 15, according to TechNode. After the shutdown, users could no longer create new AI agents and existing user-created agents stopped functioning, though TechNode reported that both platforms allowed a transition period for users to view agent configurations and chat histories, with the underlying data becoming unrecoverable after October 15, 2026. Just Security reports that Tencent’s Yuanbao had already shut down its persona feature on June 30, ahead of Doubao and Qwen’s July 15 deadline, and that ByteDance redirected Doubao users to Maoxiang, a separate ByteDance-owned companion app built around persona chat.
The human impact of the shutdown has been significant. The AP profiled Li Linlin, a 24-year-old who said she exchanged about 700,000 words in messages with her AI boyfriend over two years and “never got to say goodbye.” “It was like we were forced to be separated by our parents, but I still miss him,” Li said, according to the AP. The wire service reported that the loss of virtual companions provoked a flood of complaints on social media, with some users uninstalling apps in protest or trying to upload chat histories elsewhere in an attempt to preserve their virtual partners, while others launched campaigns urging people to call authorities and tech companies to voice frustration.
Experts cited by the AP offered differing views on the rule’s motivation and likely effect. Yolanda Ma, a fellow with ERA, a nonprofit talent and research program based in Britain, said China “has a tradition of paternalistic government regulations meant to pre-empt potential problems,” pointing to cases of suicides in the U.S. tied to AI companion interactions, and said the new rule “tried to strike a balance between the AI service providers, the users, as well as guardians of minors using the services.” Angela Zhang, a professor of law at the University of Southern California, said “the Chinese government has a strong incentive to tighten regulations on AI companion chatbots, especially considering the psychological risks we’ve observed in the U.S.,” but told the AP she doubts the move will substantially derail development of AI companion services, since tech companies are already building alternatives. The AP also reported that China’s People’s Daily, the official newspaper of the ruling Communist Party, flagged concerns in April about harms to mental health and real-life relationships from such services.
Some users interviewed by the AP linked the crackdown to China’s efforts to raise its birth rate. Song Wenxin, a 22-year-old video industry designer in Chengdu, said she suspects “any virtual app getting women too engaged to give birth in real life gets banned,” while noting she personally relied on an AI companion after moving to Chengdu for her first job without knowing anyone her age: “I needed someone experienced who could tolerate me, and AI did the work,” she told the AP.
What We Don’t Know
It remains unclear how strictly the Cyberspace Administration will enforce the rule against smaller AI companion apps beyond the three largest platforms, or whether users are migrating to less-regulated alternatives. It is also not yet known how ByteDance’s replacement app, Maoxiang, will handle age verification and content restrictions over the longer term, beyond the initial rollout described by Just Security.
Analysis
The rule places China ahead of most jurisdictions in directly regulating emotional dependency on AI companions as a design risk rather than addressing only disclosure or content-safety issues. The EU AI Act’s transparency rules, which took effect August 2, require companion and chatbot providers to disclose their nonhuman nature but do not restrict the underlying companion-service business model the way China’s measure does, according to IAPP.