Beijing adds 2 more filed generative AI services, reaching 259 models

Beijing quietly adding two more models to its generative AI filing list – taking the total to 259 – is a reminder that China’s governance strategy is about dense, continuous oversight rather than one-off flagship laws. Each filing isn’t just a bureaucratic box-tick; the rules require apps to disclose which filed model they use and to label synthetic content, which effectively ties every consumer-facing AI experience back to a monitored registry.([finance.sina.com.cn](https://finance.sina.com.cn/wm/2026-07-16/doc-inihycsp7935018.shtml))

For developers, that registry becomes a de facto whitelist: if your service isn’t on it, you are off the mainstream distribution grid in the capital’s internet economy. For established players, it raises the moat by making compliance expertise and regulator relationships part of the product. Over time, that favours large incumbents – and state-favoured upstarts – capable of maintaining multiple model variants tuned to local norms, content controls and safety expectations.

From an AGI race perspective, the filing count matters less than the infrastructure being built around it. China is standardising how models are named, audited and surfaced to users at city scale. If these registries are later linked to nationwide safety evaluations or compute quotas, they could become powerful levers for steering the direction and tempo of Chinese frontier work without having to micromanage individual labs.