China tightens child AI safety as avatar rules near

China is moving quickly to put hard edges around one of the messiest parts of the AI ecosystem: emotionally charged, anthropomorphic systems interacting with kids. The Economic Daily piece is more than a moral appeal; it is an on‑ramp for new national rules that explicitly ban virtual parents and romantic partners for minors and require youth modes and time caps for anthropomorphic services. For frontier labs, the direct impact is limited, but for companies building companion bots, virtual idols or AI tutors in the Chinese market, this is a sharp signal that business models based on parasocial attachment will face tight constraints.

In the broader race to AGI, these rules illustrate how political systems can decouple capability progress from certain kinds of deployment. China is not trying to slow down fundamental research or large‑scale training here; it is drawing red lines around particular interaction patterns that regulators view as psychologically and socially risky, especially for children. If this approach spreads, labs may be forced to cleanly separate high‑autonomy, high‑attachment use cases from general platforms, with stricter identity, age‑verification and design obligations. That would not stop AGI research, but it would narrow some of the most lucrative early consumer applications. At the same time, by normalizing sector‑specific safety regimes, China is quietly building institutional infrastructure that could later be extended to more capable systems.