Chinese AI companion services shut down after new rules take effect

Major Chinese technology platforms have pulled back popular AI companion services after new government restrictions on emotionally interactive chatbots took effect, according to an Associated Press report published August 10. The change affects services from companies including ByteDance, Alibaba and Tencent, and it is already being felt by users who had built long-running relationships with virtual characters.

The new rules, which took effect on July 15, target AI systems that simulate sustained emotional interaction through text, images, audio or video. AP reports that the regulation bars platforms from producing interactions that could manipulate users into questionable decisions, trigger extreme emotions, or encourage unhealthy habits among children and teenagers. It also requires risk warnings about overreliance on virtual companions and says replacing real-world social interaction cannot be an objective of an online service.

Why it matters

The shutdowns show how quickly companion AI can move from consumer novelty to a regulated category. Unlike general-purpose chatbots used for search, coding or productivity, companion bots are designed around persistence, personality and emotional feedback. That makes the product risk different: engagement can become dependency, and platform design choices can shape how vulnerable users respond to loneliness, stress or grief.

AP described users trying to recover chat histories, move conversations into other apps and organize complaints after losing access to virtual partners without a farewell. ByteDance, for example, is directing users of one former companion service toward another app focused on AI characters and stories, but users quoted by AP said the replacement does not feel equivalent.

The policy also places China near the front of a global debate over how to govern artificial intimacy. Regulators in several countries have raised concerns about minors, mental health, disclosure and the incentives behind emotionally persuasive AI products. China’s approach is more direct: it treats the relationship dynamic itself as a risk that platforms must limit, warn about or redesign.

For AI developers, the immediate lesson is that safety work for companion products cannot stop at content filters. Providers may need age-sensitive modes, usage limits, clearer disclosure that users are interacting with AI, crisis intervention processes, easier data controls and product designs that avoid nudging people away from human support. The Chinese shutdowns are also a warning for larger AI platforms: when an assistant becomes a companion, regulators may judge it by social and psychological effects, not just by whether individual messages violate a policy.

Sources

• apnews.com

Cover photo by Airam Dato-on on Pexels, used under the Pexels License.