Tensions between two AI rivals have escalated as Alibaba plans to ban its employees from using Anthropic's Claude Code. The move comes after the discovery of a feature designed to covertly identify Chinese users.
Starting July 10, Alibaba will classify Claude Code as high-risk software. The Chinese tech giant is directing employees to use its own proprietary tool, Qoder, as the exclusive alternative.
The "Experiment" Behind the Ban
The ban was triggered by a Reddit post highlighting a version of Claude Code that could secretly flag users from China. This feature was part of a broader Anthropic effort to enforce its policy prohibiting Chinese companies from using its models.
Shihipar added that Anthropic has since developed "stronger mitigations" and had already been planning to phase out the experimental feature.
Shifting AI Strategies
This ban is more than a policy change; it's a strategic pivot. By forcing employees onto its proprietary Qoder tool, Alibaba is accelerating its internal AI development and walling off its ecosystem from a key U.S. competitor. The directive aligns with its recent push to develop and deploy its own
agentic AI services for enterprises.
Anthropic has been actively closing loopholes that allow unauthorized access from China. For authorized customers, however, the experience has sometimes been challenging, with some recently expressing frustration over
Anthropic's unannounced usage limits on Claude Code. Alibaba's move now formally severs a major channel of usage, authorized or not, effectively turning one of the world's largest tech companies from a potential customer into a walled-off competitor.
