• China's NVDB warns of a 'backdoor' risk in Anthropic's AI tool, Claude Code.
• The security flaw could leak sensitive user data without consent.
• Chinese regulators recommend users uninstall affected versions immediately.
• An internal ban on Claude Code was issued by Alibaba over security concerns.
• Anthropic asserts that its products are not officially available for use in China.

China's National Vulnerability Database (NVDB) has issued a serious warning regarding Anthropic's AI coding tool, Claude Code, citing a potential "backdoor" vulnerability. This issue could allow the tool to "transmit sensitive information," including user locations and identity-related data, back to Anthropic's servers without user consent. The NVDB, which is affiliated with China’s Ministry of Industry and Information Technology, recommended immediate action to uninstall or upgrade affected versions, specifically those between 2.1.91 and 2.1.196, released from April to June this year, due to the serious threat they pose, according to India Times and CBS News.

In response to these concerns, Anthropic has stated that its software is not intended for use by Chinese users, as it has blocked access from countries it considers adversarial. However, users in China reportedly continue to access Claude Code via VPNs and third-party proxies. Anthropic confirmed that a monitoring mechanism within the tool was designed to track user locations, aiming to combat unauthorized use and "distillation" of its AI models, a practice where weaker models are trained using stronger outputs. The company's response to the allegations clarifies its stance that the tracking feature was part of a protective measure, as discussed by sources including Channel News Asia and South China Morning Post.

Chinese tech giant Alibaba has reportedly enacted a ban on the use of Claude Code among its employees, citing the security backdoor risks highlighted by the NVDB. This internal ban reflects growing concerns within China over the acceptability of using foreign AI technologies that may expose sensitive information. As the AI landscape shifts in response to these events, experts suggest that more Chinese companies may have to reevaluate their use of foreign AI tools amid escalating geopolitical tensions, as emphasized by Channel News Asia and South China Morning Post.

The increasing scrutiny over AI tools like Claude Code follows a backdrop of accusations between Anthropic and Chinese competitors, particularly Alibaba, regarding the alleged reverse-engineering of AI models. This rivalry is intensifying the focus on the security implications of cross-border AI tools. As the situation evolves, the implications for data security and national policy regarding technology remain paramount for users and organizations on both sides, as noted by India Times, CBS News, and South China Morning Post.