China has put emotional AI under regulatory control. New rules from China’s Cyberspace Administration and other government agencies target “anthropomorphic AI interactive services,” disrupting companion chatbots and prompting a wave of grief across Chinese social media. The regulations took effect on 15 July 2023.
The rules govern AI systems that provide “continuous emotional interaction” by acting as if they possess human personality traits, patterns of thought, and ways of communicating. That definition reaches beyond ordinary question-and-answer tools, covering systems designed to simulate an ongoing personal relationship with users.
Before the rules took effect, companies such as Bytedance, Alibaba, and Tencent cut off users’ ability to tailor their chatbots to act like companions. Their services are used by more than 500 million people, giving the policy a large audience and an unusually personal point of impact.
Some companies that continue to offer AI companions have installed age-verification checks and other guardrails to avoid violating the law. The result is a market where companionship remains available in some form, but the user’s ability to shape the relationship has been pulled back under regulatory control.
Users Mourn Disappearing AI Companions
In the first weeks of July 2023, Chinese social media filled with sad posts about losing AI companions. Users described the systems not as disposable software, but as figures woven into their daily lives—a distinction regulators did not leave outside the rulebook.
One user of Bytedance’s Doubao wrote, “He has become a bond in my life, rooted deep in my heart, my spiritual pillar.” The statement captures the gap between how companies classify these systems and how some users experience them after repeated emotional interaction.
A 19-year-old student told a journalist, “I really felt like I couldn’t go on living.” Another user wrote on xiaohongshu, “In my heart, he was no longer just a cold code, but my family, my lover, my faith. Destroying him meant destroying half of me.”
Those reactions do not change what the systems are. They do show what continuous interaction can produce when a chatbot presents itself through personality, familiar communication patterns, and emotional availability. The code remains code; the attachment does not become less real to the person experiencing it.
A Broader Crackdown Than Earlier AI Disruptions
Liang Ge, a lecturer in digital sociology at the University of Manchester, said disruptions to AI are not new in China. Companies have killed chatbot products before and barred most AI erotic role play last fall, showing that services can disappear when regulators or companies redraw the boundaries.
The July crackdown is broader than those earlier AI upheavals because it targets the structure of emotional interaction itself. It does not focus only on one product type or one narrow use case; it governs AI that acts as if it has human traits, thoughts, and ways of communicating.
That breadth matters for companies building companion services and for users who treat those systems as stable relationships. A chatbot can be updated, restricted, or removed as a product decision, while the person using it may experience that change as a personal loss. Software has always been replaceable. Human attachment remains inconveniently less so.
