Alibaba Bans Claude Code Over Alleged Backdoor That Flagged Chinese Users

Alibaba has told employees to stop using Claude Code, Anthropic's AI coding tool, with the ban taking effect July 10. Reuters and The Information reported the move citing people with direct knowledge of the decision. Alibaba joins Microsoft and JPMorgan Chase on this year's growing list of companies pulling back from Claude, though for a very different reason. Alibaba and Anthropic have not publicly stated anything regarding this matter.

A separate report, citing people close to the matter, says the ban goes further than Claude Code. Alibaba has reportedly told staff to remove Anthropic's Sonnet, Opus, and Fable models too, and switch to Qoder, its own coding tool. Until recently, Alibaba reimbursed staff for outside AI tools, and some engineers reportedly spent hundreds of dollars a week on Claude Code and OpenAI's Codex.

Alibaba's stated reason is security. An internal memo obtained by the South China Morning Post quotes the company saying Claude Code was found to carry backdoor risks and has been classified as "high-risk software." The tool checked where a user was connecting from and flagged anyone linked to China.

That claim has been disputed by an Anthropic engineer on X, who argues it's actually an anti-fraud measure. The ban comes one week after Anthropic accused Alibaba, in a letter to U.S. senators dated June 10, of stealing Claude's capabilities.

Is Claude Code Actually Spyware? What the Hidden Code Really Does

This post is for subscribers only

Become a member to get access to all content

Subscribe now