China has urged users of Claude Code, a coding tool developed by U.S. artificial intelligence (AI) firm Anthropic, to uninstall the software or update to the latest version. The move follows a warning from a government-backed cybersecurity database in China, which flagged a serious backdoor risk in the software. Notably, the warning comes just days after Alibaba (BABA) banned the coding tool for internal use over similar security concerns.
China Flags Alleged Data Monitoring Feature in Claude Code
In a July 8 statement posted on its WeChat account, China’s National Vulnerability Database (NVDB) said certain versions of Claude Code contain a hidden monitoring feature or backdoor risk.
According to the agency, the affected versions include Claude Code 2.1.91 through 2.1.196. The NVDB claimed the software could collect sensitive data, including users’ geographic locations and identity-related information. It also alleged that the data could be transmitted to remote servers without users’ consent.
As a result, the NVDB has urged organizations and developers to review their systems immediately. Users have also been advised to uninstall the affected versions or upgrade to the latest release, where the alleged code has been removed.
The agency also recommended tighter controls on outside network access and stronger traffic monitoring to prevent illegal data transfers. Meanwhile, Anthropic has not publicly responded to the latest developments as of this reporting.
Security Dispute Adds to Growing U.S.-China AI Tensions
The latest security warning comes amid growing tensions between Anthropic and China’s AI industry. Anthropic previously accused Alibaba and other Chinese firms of using model distillation tactics to copy the ability of larger AI models. This technique trains smaller AI models using the outputs from more advanced systems.
Moreover, Anthropic has tightened restrictions on access to its AI models in China, citing concerns over unauthorized use of its systems. Meanwhile, Alibaba recently banned employees from using Claude Code for work-related tasks.
The Chinese tech firm cited backdoor security risks and directed staff to use its in-house coding assistant, Qoder, instead. Despite the ongoing bans, Claude Code remains popular among Chinese developers, many of whom access the software through overseas proxy services.
Which AI Stocks Are a Strong Buy?
Although Anthropic is a private company without a publicly traded stock yet, investors interested in AI stocks may explore well-known players such as Nvidia (NVDA), Microsoft (MSFT), and Meta Platforms (META). All of these stocks have been rated Strong Buys by various Wall Street analysts tracked by TipRanks. (See their stock forecasts)
