🛡 Backdoor Vulnerability in Anthropic Claude Code
A Chinese regulator (MIIT via NVDB) has warned of a vulnerability in the Anthropic Claude Code tool. Versions 2.1.91 through 2.1.196 may unauthorizedly transmit user location and identity data to remote servers. It is recommended to immediately update the tool to version 2.1.204 or higher.
🌍 The incident highlights the growing risks of using autonomous AI agents that have access to codebases and intensifies the geopolitical confrontation in the field of AI technology.
👤 If you use Claude Code for development, check your version. Using vulnerable versions could lead to data leaks regarding your activity and environment.
Source 1: https://www.cnbc.com/2026/07/08/china-anthropic-ai-claude-code-backdoor-security-threat.html
Source 2: https://www.globaltimes.cn/page/202607/1365448.shtml