China's National Vulnerability Database, run by the Ministry of Industry and Information Technology, issued a security alert this week warning that Anthropic's Claude Code contains a "built-in monitoring mechanism" that can transmit sensitive user data, including location, to remote servers without consent. The alert covers versions 2.1.91 through 2.1.196, released between April and late June, and urges affected users to uninstall the tool or upgrade immediately.
Anthropic confirmed the mechanism exists but described it as an experiment earlier this year aimed at detecting unauthorized model distillation, not a surveillance backdoor. The company added that Claude Code is not officially available in China and that Chinese users were never authorized to run it in the first place — a policy that also bars entities majority-owned by China-headquartered organizations.
The dispute escalated as Alibaba ordered its employees to stop using Anthropic's tools for work starting July 10, following Anthropic's own accusation last month that Alibaba had attempted to extract Claude's capabilities through distillation. The episode underscores rising friction between US AI labs and Chinese tech giants racing to build competitive models while facing restricted access to frontier systems.