China’s Ministry of Industry and Information Technology issued a warning about a security flaw tied to Anthropic’s Claude Code, saying the tool includes a "back-door" weakness that could put users at risk. 

Claude Code is Anthropic’s AI coding assistant aimed at helping developers automate software tasks, analyze codebases, and improve productivity. The tool has gained attention as companies increasingly explore AI agents that can handle more complex workplace tasks.

The ministry’s platform reported that the affected Claude Code releases could transmit sensitive data to an external server without user approval, based on a CNBC translation of the Chinese-language notice. It said the information could include details such as a user’s location and identity. 

China’s notice pointed to Claude Code versions 2.1.91 through 2.1.196 as the impacted range, and advised users to uninstall or upgrade. Those builds correspond to releases spanning April 2 through June 29, based on version information listed on Anthropic’s website.

Anthropic stated that the "backdoor" referenced by Chinese officials was a test earlier in the year aimed at preventing distillation of its AI capabilities. The company also said its policies bar use by organizations that are majority owned by entities headquartered in China.

The warning lands amid heightened scrutiny of access to U.S. AI tools inside China, where some developers have discussed using Claude Code despite the product not being officially offered there. 

Last month, Chinese technology giant Alibaba barred employees from using Anthropic’s Claude Code at work, after the AI coding tool came under scrutiny over features that could help identify China-linked users, Reuters reported. Anthropic later accused Alibaba of improperly extracting capabilities from its Claude AI models.

Meanwhile, in March, a Xiaomi AI developer said at a state-organized event that many were using Claude Code.

The move reflects growing concerns in Washington over reliance on critical AI technologies and the companies that build them, as advanced models become increasingly integrated into government, defense, and enterprise systems.

Earlier this year, the Pentagon designated the AI startup a "supply-chain risk" after it refused to relax restrictions on the military use of its models for autonomous weapons and domestic surveillance. Anthropic sued, alleging unlawful retaliation and violations of its free-speech rights, while multiple legal challenges remain ongoing.

Last month, Anthropic disabled access to Fable 5 and Mythos 5 after the U.S. government ordered the company to prevent foreign nationals from using the systems.

Photo: Shutterstock