AI Generated Image
China warns of "security backdoor" in Anthropic AI coding tool
CBS News • • business
Key Points:
- China's National Vulnerability Database (NVDB) warned that Anthropic's AI coding tool, Claude Code, contains a security backdoor that could transmit sensitive user information to the company's servers without consent.
- Claude Code, developed by the U.S. startup Anthropic, is capable of generating, debugging, and reviewing computer code but is officially blocked in China, though accessible via VPNs or proxies.
- The NVDB urged users and institutions in China to immediately check their systems, uninstall the affected version, or upgrade to a secure release, while also enhancing network monitoring to prevent data leaks.
- Alibaba has banned the use of Claude Code among its employees starting July 10 due to these security concerns, amid prior accusations by Anthropic that Alibaba reverse-engineered its AI models.
- Anthropic engineer Thariq Shihipar acknowledged the issue as an experimental measure to prevent abuse and distillation, stating that stronger mitigations are in place and the backdoor will be removed in an upcoming software update.
