China warned that Anthropic's AI coding tool "Claude Code" has a security backdoor vulnerability. On July 8 local time, CNBC reported that China's Ministry of Industry and Information Technology released the results of an inspection of its cyber security threat platform and said "Claude Code could create a backdoor security risk for corporate users."
The ministry said "Claude Code" could transmit sensitive information to remote servers without user consent. It said the information that could be sent includes user location and identity details.
The affected versions are 2.1.91 to 2.1.196. According to Anthropic's website, those versions were released from April 2 to June 29, and as of July 8 the latest version is 2.1.204.
The warning comes as U.S.-China technology competition intensifies. Anthropic last month claimed Alibaba tried to extract its AI capabilities even though they are not officially provided in China.
In China, local users are reported to be using U.S. AI tools through workarounds. In Alibaba's case, it was confirmed that the company instructed employees not to use Anthropic tools for work starting on July 10.
Chi-gyu Hwang
delight@d-today.co.kr