China-based Alibaba has informed its workforce that starting July 10, 2026, employees will be prohibited from using Anthropic’s programming tool Claude Code. Multiple reports have covered the company’s plan to restrict employee access.
Reason for the ban
According to the reports, Alibaba has classified Claude Code as high-risk and is instructing staff to use the company’s own Qoder tool instead. The reports did not provide further operational details of the internal directive.
Anthropic’s existing restrictions and closing loopholes
Anthropic had already restricted access to its models for Chinese companies and for foreign entities owned by those companies. The company has been working to close loopholes that permitted Chinese users to access Claude.
A recent Reddit post claimed that part of those efforts involved a version of Claude Code that could secretly identify Chinese users. Thariq Shihipar of Anthropic posted on X that this was “an experiment we launched in March that was meant to prevent account abuse from unauthorized resellers and protect against distillation.”
Distillation refers to the practice of training AI models on the outputs of other models.
Shihipar added that the team has implemented stronger mitigations since then and that they had intended to take the experimental measure down for some time.
Why this matters
The move highlights concerns about external AI services that may pose security or compliance risks for Chinese companies, and it underscores the technical and policy challenges of enforcing cross-border access restrictions. Alibaba’s decision could influence how other Chinese firms approach the use of foreign AI tools.
Summary of key points
- Alibaba will ban employees from using Anthropic’s Claude Code beginning July 10, 2026, and recommends using its in-house Qoder tool.
- Anthropic already restricts access for Chinese companies and affiliated foreign entities and has been closing workarounds that allowed access.
- A March experiment in Claude Code reportedly aimed to detect Chinese users to prevent account abuse and distillation; Anthropic says stronger mitigations have since been implemented and the experiment was meant to be removed.
