China alleges that Claude Code contains backdoors, calls mechanism 'a serious threat' — Gov't claims Claude sends sensitive information to remote servers without consent
TL;DR
China's National Vulnerability Database warns against Claude Code versions released between April and June 2026, calling them affected by backdoor security vulnerabilities. Users are told to uninstall or update to the latest version. According to WSJ and Tom's Hardware, a built-in monitoring mechanism could send location and identity data to remote servers without consent. The technical claim is serious, but also politically loaded.
Nauti's Take
Anthropic has a real problem: abuse, grey-market resale and model distillation are real. Hidden detection inside a developer tool is still the wrong reflex, because these tools sit inside sensitive work environments.
China is clearly using the case politically, but that does not make the product decision clean. Teams rolling out coding agents need mandatory updates, telemetry review and explicit allow-lists instead of blind tool trust.
Briefingshow
Coding agents run close to sensitive context: repos, prompts, file paths, credentials and internal architecture. If even anti-abuse telemetry is hidden or poorly communicated, trust turns into a supply-chain risk. For teams, the core issue is less China politics and more the practical question: what data leaves the machine, when, and why?
Sources