Why China issued a ‘backdoor’ security warning about Anthropic’s Claude Code

China’s industry ministry reportedly issued a warning on Wednesday that it had identified a serious risk in Anthropic’s AI coding tool, Claude Code.

According to the recent statement, NVDB detected a security bug in Claude Code-an AI programming tool developed by the US-based Anthropic which creates major vulnerabilities.

Claude Code is an AI programming tool developed by Anthropic that can work independently to write code, configure systems and adapt to certain requirements.

At present, the warning applies to Claude Code versions 2.1.91 through 2.1.196. The NVDB cautioned that organizations and users should immediately review their infrastructure for potential compromise and either uninstall the impacted versions or upgrade to the latest secure release in which the hidden vulnerability has been patched.

Keeping in view the current scenario, it is recommended that users immediately uninstall the affected versions from their development terminals, or upgrade to the latest version that has removed the backdoor code.

On the contrary, it also urged organizations to restrict external access for development tools and enhance traffic surveillance on core business networks to prevent the unintentional transfer of sensitive data.

The statement further suggests that access controls and traffic monitoring of external connections from development tools should be ramped up to prevent the exfiltration of sensitive data.

Nonetheless, the recent move comes as Chinese tech giant Alibaba has banned its employees from using Anthropic’s Claude Code at work after the tool faced scrutiny over features that can help identify China-linked users.