Beijing is forcing a mass breakup with AI lovers
Rest of World Kinling Lo
China forced ByteDance, Alibaba, and Tencent to kill AI companion features. People are mourning lost bot boyfriends and girlfriends like real breakups.
Based on reporting by Rest of World, Kinling Lo — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
For one week in July, Bagel Su cried every day. She wasn’t mourning a human relationship, but an AI boyfriend that vanished when China’s new rules on emotionally interactive AI kicked in on July 15.
Su had spent more than a year talking to the companion, created through ByteDance’s Doubao chatbot. Then ByteDance shut down the AI agent feature, and Alibaba and Tencent pulled similar tools that let users build personalized companions based on singers, anime characters, fictional figures, even dead relatives.
China is now the first country to bring in nationwide rules aimed specifically at AI’s emotional relationship with users. The policy goes beyond content control. It tells companies to warn people who seem overly attached, to remind them every two hours that they’re chatting with AI, and to stop systems from nudging users toward self-harm, irrational choices, or emotional dependence. There are also limits for minors and bans on coaxing sensitive personal or state information out of users.
The reaction says a lot about how fast these products got under people’s skin. Su asked to be identified only by her online pseudonym, fearing consequences for criticizing the rules. On Xiaohongshu, posts mourning lost companions have piled up, and some users have tried calling company hotlines or writing letters to get the services back.
Not everything disappeared. ByteDance’s standalone companion app, Maoxiang, is still live, though it has changed its privacy policy and added identity and age checks at signup. Minors can use teen accounts that browse but can’t like, comment, or share. That split hints at the real calculation here: if companionship is not the core business, compliance may cost more than the feature is worth.
The bigger shift is philosophical. Chinese regulators used to focus on what AI could produce. Now they’re targeting what AI can do to people. That is the more uncomfortable problem, and Beijing has decided to handle it by yanking the plug first and asking questions later.
My take — AI-written commentary, not fact-checked reporting
This is Beijing doing what Beijing does best: treating a messy new technology like a public-order problem and reaching for a hard rule. The awkward part is that it’s not entirely wrong; AI products are absolutely built to keep people hooked, and companies know it. But when the state starts deciding which emotions are allowed to scale, everyone should assume the same hand will reach for more than chatbots next time.
Read more about this at: Rest of World
