China bans AI chatbot personas and mandates neutral virtual assistants

The Cyberspace Administration of China (CAC) has ordered major Chinese tech firms—including Baidu, Alibaba, ByteDance and Tencent—to remove all personality‑style features from their AI chatbots. The directive, which takes effect on 15 July, requires virtual assistants to provide neutral, fact‑based responses and prohibits the offering of anthropomorphic AI services to minors without parental consent.

Companies have already disabled the customizable persona functions of products such as ByteDance’s Doubao, Alibaba’s Qwen and Baidu’s Ernie Bot. The move is presented as a safeguard against misinformation, “AI psychosis” and the creation of virtual intimate relationships that could mislead users. The CAC cited that “excessive personalization can lead to disinformation and loss of trust in automated systems.”

The regulation is part of a broader Chinese strategy to shape its AI industry, emphasizing control and social stability while competing globally with cheaper, high‑capacity models. It signals a tightening of the domestic AI market and may affect international developers that partner with Chinese platforms.