China publishes 15th Five-Year Plan for National Informatization: names AI law and extends model filing to AI agents

On 23 October 2026, the Central Cyberspace Affairs Commission (CAC) published China's '15th Five-Year Plan for National Informatization' (2026-2030), a 14-page master plan for China's digital build-out. It names 'advance foundational legislation on artificial intelligence' as a formal work item (corroborating the State Council 2026 legislative work plan and the 15th FYP outline), commits to 'optimize the generative AI filing mechanism', and explicitly calls for establishing security risk assessment and filing for new forms such as AI agents, on-device large models and anthropomorphic interaction ('companion') services. It also directs research on judicial interpretations for AI, reinforces cross-border data facilitation, and promotes a national network ID.

This is the authoritative Chinese state policy signal that the AI governance regime is about to extend beyond single models to agentic and on-device AI, and that a comprehensive AI law is coming. AI providers serving the Chinese market (including those selling both to China and the EU) should anticipate CAC filing/assessment requirements for agentic systems and companion applications, alongside the EU AI Act's high-risk registration obligations. It sets the compliance trajectory for the next five years.

AI providers with China market exposure should monitor CAC draft rules on agent/on-device/companion model filing and on the 'optimisation' of generative AI filing; begin mapping AI-agent and companion-service data flows and assessment readiness ahead of expected CAC drafts.