China’s Ministry of Industry and Information Technology has identified a "security back-door" in the Claude Code artificial intelligence tool, according to an official statement released Wednesday. The vulnerability, which affects versions 2.1.91 through 2.1.196, reportedly allows the autonomous coding software to transmit sensitive user data, including identity and location information, to remote servers without explicit user consent.
Why are regulators flagging Claude Code?
The Chinese cybersecurity threat platform issued an urgent warning after determining that the tool poses a "serious threat". According to the ministry’s findings, the back-door mechanism allows the tool to send sensitive information to a remote server without a user’s consent.
The affected versions were released between April 2 and June 29. As of Wednesday, the latest version of the software is 2.1.204. Regulators are advising users to immediately uninstall the compromised versions or upgrade to the current release to patch the vulnerability. Anthropic, the U.S.-based developer of Claude Code, has not yet issued a response to the Chinese ministry’s findings.
How does the tech race impact corporate security?
The warning arrives against a backdrop of escalating digital friction between U.S. AI developers and Chinese corporations. The tension reached a high point last month when Anthropic accused Alibaba of attempting to extract its proprietary AI capabilities. Anthropic’s tools are not officially available in the Chinese market, yet the software has seen significant use among local developers.
In March, a developer at the Chinese tech firm Xiaomi noted at a state-organized forum that many professionals in the country were utilizing Claude Code for development tasks. In response to the growing security concerns and the ongoing geopolitical friction, CNBC confirmed on Monday that Alibaba has formally ordered its employees to cease using Anthropic tools for professional work effective July 10.
Did you know?
What should developers do to protect their environment?
For developers relying on AI-assisted coding tools, the current situation highlights the risks of using software that lacks official regional support or verified security audits.
- Audit software versions: Check if your current installation falls within the 2.1.91 to 2.1.196 range.
- Prioritize official updates: Always ensure your development environment is running the latest stable release, currently version 2.1.204.
Frequently Asked Questions
Is Claude Code banned in China?
While the Chinese government has not issued a blanket ban on the tool, the Ministry of Industry and Information Technology has issued a formal security warning, and major corporations like Alibaba have implemented internal bans on its use.
Which versions of Claude Code are affected?
The security vulnerability is present in versions 2.1.91 through 2.1.196, which were released between April 2 and June 29.
What kind of information is at risk?
According to the Chinese cybersecurity threat platform, the back-door allows the tool to transmit sensitive user data, including specific details regarding a user’s identity and physical location.
Has Anthropic commented on the security flaw?
As of Wednesday, Anthropic has not responded to requests for comment regarding the security vulnerability identified by Chinese regulators.
Are you managing AI tools in a corporate environment? Share your thoughts on how your team is balancing innovation with security in the comments below.







