Claude Code: China Warns of Backdoor Risk in specific versions of the AI coding tool, urging organizations to remove them immediately due to alleged security vulnerabilities. The warning, published by Chinese authorities, claims that certain Claude Code versions contain a backdoor that could be exploited by attackers. However, Anthropic, the developer of Claude Code, has rejected this claim, stating that the feature in question is actually an anti-abuse protection mechanism. The dispute highlights growing tensions between global tech companies and cybersecurity regulators.
The Chinese government’s alert specifically targeted older Claude Code builds, advising domestic firms to delete these versions and update to the latest release. Officials described the alleged backdoor as a potential entry point for malicious actors to compromise systems. This move is part of China’s broader effort to tighten control over foreign software tools used within its borders, particularly those involving AI and code generation.
Anthropic responded by clarifying that the flagged feature is designed to detect and block abusive usage of Claude Code. The company emphasized that it takes security seriously and that the functionality is standard for preventing misuse of the tool. Anthropic urged users to rely on official updates and dismissed the backdoor accusation as a misunderstanding of the anti-abuse system.
The incident serves as a reminder of the challenges faced by international tech firms operating in China. It also underscores the importance of keeping software updated and verifying security warnings from official sources. Users of Claude Code should ensure they are running the latest version to avoid potential risks and compliance issues.