China and the United States are escalating warnings related to AI developer tools, according to reports from both sides. China’s Ministry of Industry and Information Technology (MIIT) advises that Anthropic’s Claude Code should not be used, stating that a Chinese cybersecurity platform has identified “backdoor” risks associated with the agentic coding tool. MIIT’s position is described as a broader warning to companies to remove or stop using Claude Code in China.
Anthropic responds by disputing the concern. The company says that users in China being told to uninstall Claude Code are not supposed to be using the product, and it frames MIIT’s warnings as a mismatch between authorized use and how the tool is being used.
The dispute is unfolding alongside a wider US-China AI policy divide. US lawmakers are reported to be moving against American firms that rely on low-cost Chinese AI models, while China simultaneously increases scrutiny of US AI software and its security implications. The exchanges reflect growing regulatory and security tensions around AI systems and their access, provenance, and safeguards.
