Alibaba Bans Anthropic’s Claude Code Over Hidden Tracking Feature and Security Concerns
Alibaba Group has barred employees from using Anthropic’s AI coding assistant Claude Code for work, citing security concerns after the US startup acknowledged embedding hidden code that monitored certain user environments.
According to an internal notice, the Chinese technology giant said, “As Claude Code was recently discovered to carry back-door risks, after comprehensive evaluation, Claude Code has now been added to a list of high-risk software with security vulnerabilities.”
The ban will take effect from July 10, prohibiting employees from using the coding tool within the workplace. It should be noted that Alibaba has instead directed staff to use its in-house AI coding platform, Qoder.
Tracking Feature Fuels Security and Trust Concerns
The decision follows revelations that Anthropic had embedded code in Claude Code designed to identify whether users were based in China or linked to Chinese AI laboratories.
Security researchers uncovered the feature earlier this week on platforms including Reddit and GitHub, with reports claiming the software collected details such as users’ time zones, proxy usage, and other environmental information before sending it back to Anthropic.
Responding on X, Anthropic engineer Thariq Shihipar said the tracking mechanism was “an experiment” introduced in March to “prevent account abuse from unauthorised resellers and protect against distillation.”
He added that the feature would be removed as part of a software re-release because the company had developed “stronger mitigations”.
The controversy prompted criticism from cybersecurity experts. Beijing-based developer Huang Yong said that secretly inserting such code effectively created potential “back doors” on developers’ computers, while Chinese cybersecurity firm Huorong Security warned the practice also raised cross-border data compliance concerns.
Google – Gemini
Latest Flashpoint in the US-China AI Rivalry
Alibaba’s move comes amid escalating tensions between US and Chinese AI firms. The ban follows Anthropic’s recent accusations that Alibaba attempted AI model “distillation”, the practice of training a smaller model using outputs generated by a more advanced one.
Anthropic made the allegation in a letter to two US senators, claiming such efforts could accelerate China’s AI capabilities.
Despite Anthropic officially restricting access to its AI models in China, Claude Code has remained popular among Chinese developers through various workarounds.
Analysts say the latest episode underscores how AI competition is increasingly extending beyond technological leadership into cybersecurity, compliance and digital sovereignty.
As concerns over legal risks and access controls grow, Chinese firms are increasingly shifting toward domestic AI ecosystems, including Alibaba’s Qwen, DeepSeek, Moonshot and Zhipu, reducing dependence on US-developed AI tools.
