Alibaba Bans Claude Code Over Tracking Concerns

Amid intensifying competition in the artificial intelligence (AI) industry, Chinese e-commerce giant Alibaba has banned the use of Claude Code, the AI-powered coding tool developed by U.S.-based AI company Anthropic. From July 10, Alibaba employees will no longer be permitted to use the software, according to multiple international media reports.

In an internal notice, Alibaba classified Claude Code as "high-risk software" and instructed employees to remove it from their computers. The company has instead recommended that staff use its in-house AI coding assistant, Coder.

The decision follows allegations that surfaced on June 30 in a post on the social media platform and online forum Reddit. In the post, a technology researcher claimed that an analysis of Claude Code's source code revealed a concealed tracking mechanism embedded in the software. According to the allegation, the feature could determine whether a user was located in China or affiliated with Chinese AI laboratories such as Alibaba, Baidu, or ByteDance by examining the system's time zone and internet proxy settings.

The allegation further claimed that Anthropic did not transmit the tracking data directly. Instead, it allegedly employed a form of steganography—hiding information through subtle, nearly imperceptible changes in elements such as system prompt dates or punctuation—to send signals back to servers in the United States.

Following the disclosure, Tharik Shihipar, an engineer on Anthropic's Claude Code team, acknowledged the existence of the mechanism in a post on X. However, he described it as an "experimental project," saying, "We launched this experiment last March. Its purpose was to prevent unauthorized resale of our accounts on the black market and to protect our AI models from distillation." He added that the code had already been removed from the software.

The report comes against the backdrop of existing U.S. restrictions on Chinese companies' access to Anthropic's AI models, imposed on national security grounds. Despite those restrictions, many Chinese researchers reportedly continued using Claude through proxy services and other workarounds. In June, Anthropic also alleged that an Alibaba laboratory had created approximately 25,000 fake accounts to extract data from Claude.

Industry observers speculate that the alleged tracking mechanism was developed in response to that dispute and that Alibaba's decision to prohibit the U.S.-developed software is a direct countermeasure.

//DBTech/BMT/OR//