China imposes new rules on AI companion services

Effective 15 July 2026, China’s Cyberspace Administration announced sweeping regulations for generative‑AI companion products that simulate human personalities. The measures ban emotional‑AI features for users under 18 and impose strict age‑verification, content‑filtering, data‑localisation and transparency requirements on services aimed at adults. Companies must label AI interactions, limit sessions to two hours, and undergo security reviews once they reach one million registered users or 100 000 active monthly users. The rules also oblige providers to monitor signs of affective dependence, mental‑health distress and unsafe behaviour, and to cease using private conversations to train future models.

The policy aims to protect national security, public interest and citizens’ rights, citing concerns that “affective‑AI dependence could harm mental health and even affect demographic trends”. Analysts note the compliance burden will push smaller startups out and accelerate consolidation among larger tech firms, reshaping China’s fast‑growing AI‑companion market.