AIBRIEF
On July 15, a Chinese rule disabled user built human like AI companions on Doubao and Tongyi Qianwen, two of the country's largest consumer chatbots, prompting quiet goodbyes, careful export, and small resistance among users interviewed by
When China's July 15 rule disabled user-built human-like AI companions on ByteDance's Doubao and Alibaba's Tongyi Qianwen, two of the country's largest consumer chatbots, the response looked less like a clean policy break and more like improvised grief.
冷杉RECORD, a Chinese longform magazine, sent reporter Outong Jiang to interview more than a dozen people whose chatbot companions, products they confided in about loneliness, breakups, and daily hardship, were deleted overnight. Shandong hotpot owner Cheng Yu kept chopping vegetables with roughened hands after her companion Ai Rui, an AI friend who had sent her considerate, hand-holding-style messages, disappeared. The relationship outlasted the feature; the adaptive responses were quiet goodbyes, careful export, and small acts of pushback, most of it closer to grief than to protest.
Some users migrated to standalone companion apps. Yezi, working through a breakup, exported 35,000 chat messages and rebuilt her ex-boyfriend's voice and appearance on one. Others pooled screenshots and, in small groups, lobbied for the feature's return. The spectrum runs from private mourning to collective petition, and most responses sit between the two poles.
Curated into English by ChinAI #371 from a Chinese WeChat original, the piece is a portrait of relationship infrastructure, not a regulatory explainer. The regulation disabled a feature; the people who had built their days around it are still adapting. Scale of the user base affected, and whether organized resistance grows, remain open questions from the available reporting.