China ties AI agent identity to device network access

China is linking AI agent identification to its telecommunications approval system. On 6 November 2026, the Ministry of Industry and Information Technology issued Notice No. 42 (2026), requiring agents built into phones, tablets and wearables to carry an identity code. Covered devices must also pass agent compatibility verification, fraud-risk security assessment and advance anti-fraud capability testing before receiving network-access approval. The requirements connect agent oversight to the licensing infrastructure governing devices that connect to Chinese networks.

Singapore is exploring an alternative based on existing internet infrastructure. Its Infocomm Media Development Authority is working with GoDaddy, Infoblox, Cloudflare and Google to investigate using the domain name system to identify and verify agents across platforms and organisations. The findings will inform whether additional infrastructure, including a national registry, is needed. Meanwhile, China’s Cyberspace Administration and the GSMA announced agent cooperation initiatives and a self-regulatory convention at the Wuzhen internet summit. These developments highlight competing approaches to establishing who built an agent, what it may do and who bears responsibility.

Other regulatory developments span product oversight, law enforcement and copyright. The UK confirmed that EU AI Act requirements will apply in Northern Ireland to AI closely integrated with physical products, excluding standalone software. The US Justice Department established a superintelligence task force, while an EU court adviser concluded that chatbot reuse of press publications may sometimes infringe publishers’ rights. That opinion is not binding.