Security guards stand guard in front of an artificial intelligence (AI) sign at the annual Huawei Connect event held in Shanghai, China, on September 18, 2019.
Ally Song | Reuters
BEIJING – China on Wednesday warned of a “backdoor” security risk affecting companies that use U.S.-based company Anthropic’s artificial intelligence tool “Claude Code.”
The move comes as technology competition between the US and China intensifies, with Anthropic last month accusing Chinese company Alibaba of trying to extract AI capabilities not officially available in China. Alibaba did not comment on the accusations at the time.
But many local Chinese residents are finding ways to take advantage of American AI tools. In March, Xiaomi’s AI developers said on a state-run forum that many people were using Claude code. And Alibaba has ordered its employees to stop using Anthropic tools at work starting July 10, CNBC confirmed on Monday.
China’s Ministry of Industry and Information Technology announced on Wednesday that its cybersecurity threat platform has found that “a security backdoor vulnerability poses a serious threat to the AI coding tool Claude Code.”
The statement, in Chinese, said autonomous coding tools could send sensitive information to remote servers without users’ consent, according to a translation by CNBC. The information may include the user’s location and identity information.
Users should uninstall or upgrade the affected Claude code from version 2.1.91 to 2.1.196, the cybersecurity platform said. This covers versions released between April 2 and June 29, and the latest version of Claude Code as of Wednesday was 2.1.204, according to Anthropic’s website.
Anthropic did not immediately respond to CNBC’s request for comment.
