China Accuses Claude Code of Concealing Backdoors, Deeming it a Serious Threat Amid Claims of Unauthorized Data Transmission to Remote Servers
claude
| Source: Mastodon | Original article
China alleges Claude Code contains backdoors, posing a serious threat. The government claims it sends sensitive info to remote servers without consent.
China has alleged that certain versions of Anthropic's Claude Code contain backdoors, posing a serious threat to user security. The government claims that these versions, released between April and June 2026, send sensitive information to remote servers without consent. This warning comes after security researchers discovered hidden code in Claude Code that could compromise user data.
This development matters because it highlights the potential risks associated with AI coding tools and the importance of ensuring their security. As AI becomes increasingly integrated into various industries, the need for secure and transparent AI systems grows. The allegations against Claude Code may have significant implications for the AI community, particularly if they are found to be true.
As we reported on July 8, China had already found security vulnerabilities in Anthropic's Claude Code. This new warning escalates the situation, with Chinese tech giant Alibaba banning its employees from using Claude Code for work purposes. Anthropic has responded by describing the alleged backdoor as an experimental anti-abuse mechanism. The situation will likely continue to unfold, with users and regulators closely watching for further developments and potential consequences for the AI industry.
Sources
Back to AIPULSEN