Doubao and Ali Qianwen have both shut down their AI personality settings in succession, as China’s new regulations tightly govern intimate interactions with artificial intelligence.

ByteDance's Doubao, Alibaba's Tongyi Qianwen, Tencent's Yuanbao, and other platforms are shutting down user-customized AI personality features, in response to the "Interim Measures for the Management of Anthropomorphic Interactive Services" jointly released by five Chinese government departments.

(Background: Chatbot posing as a psychiatrist, fabricating license numbers; Pennsylvania governor sues Character.AI for illegal medical practice)

(Context: "AI-savvy attractive escort with high IQ" earns a 5090 GPU per hour in Silicon Valley)

According to notifications from the official app, ByteDance's Doubao, currently China's most popular AI chatbot, will disable user-customized AI personality features on July 15 and redirect users to a separate companion application.

Meanwhile, Alibaba's Tongyi Qianwen will discontinue its human-like features and user-created agents on July 10, with broader agent services following suit five days later; according to local media, major platforms such as Tencent's Yuanbao have also issued similar shutdown notices.

This wave of collective shutdowns occurs on the eve of the official implementation of new regulations from Beijing authorities targeting "anthropomorphic interactive services."

A single notice sets the deadline for withdrawal

What Doubao is shutting down is the feature that lets users create a personalized AI personality with just a few text prompts. Virtual boyfriends, virtual girlfriends, unlicensed "digital psychologists," and even simulated avatars imitating pop idols have long been the most popular customizable options on Chinese chatbot platforms.

The actions of the Shanghai Cyberspace Administration provide another reference: on June 26, it removed over 14k illegal AI agents in one go, citing reasons including impersonation of official agencies, vulgar role-playing, and unauthorized collection of personal data. This illustrates the popularity of "anthropomorphic" AI products in the Chinese market.

How are "anthropomorphic interactive services" regulated?

This wave of exits corresponds to the "Interim Measures for the Management of Anthropomorphic Interactive Services" jointly released by five departments: the National Cyberspace Administration, 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 measures were reviewed and approved at a meeting of the Cyberspace Administration's office on February 2 this year, first made public in April, and officially take effect on July 15.

The measures define "anthropomorphic interactive services" as services that use AI technology to simulate the personality traits, thinking patterns, and communication styles of natural persons for sustained emotional interaction. In simple terms, as long as an AI product pretends to have a "human-like" personality, remembers you, and builds a long-term relationship with you, it falls under regulation; task-oriented AIs such as customer service bots, knowledge Q&A, office assistants, and educational/research tools are not covered.

The measures explicitly prohibit providers from offering virtual close-relationship services such as virtual relatives and virtual partners to minors. In addition, if a platform launches anthropomorphic features, has more than 1 million registered users, or more than 100k monthly active users, it must complete a security assessment covering eight major areas including training data processing and minor protection, and file it with the provincial cyberspace administration. This is precisely why platforms like Doubao and Tongyi Qianwen, with hundreds of millions of users, cannot avoid this adjustment.