China’s Alibaba Group will stop employees from using Anthropic’s Claude Code starting on July 10, according to multiple reports. The tech giant classifies the programming tool as high-risk software and directs staff to switch to its own internal alternative, Qoder. This move follows existing restrictions that already prevent Chinese companies and foreign entities owned by them from accessing Anthropic’s models. Reports indicate that Anthropic previously identified loopholes allowing Chinese users to bypass these blocks. One such method involved a version of Claude Code capable of secretly identifying users from the region. Thariq Shihipar, a representative from the company, described the feature as an experiment launched in March to prevent account abuse by unauthorized resellers and protect against model distillation. Shihipar noted that stronger mitigations are now in place and the tool has been ready for removal for some time. Despite these internal efforts to close gaps, the ban remains a significant operational shift for Alibaba’s development teams.
The decision highlights the growing friction between global AI tools and local data sovereignty concerns. It forces developers within the region to rely on domestic infrastructure, potentially limiting access to specific code generation capabilities. The incident also reveals how companies attempt to enforce geographic boundaries through technical detection methods.
- Employees must switch to Alibaba’s Qoder tool immediately.
- Anthropic previously detected Chinese users via secret identification features.
- The ban aligns with existing policies restricting Chinese entities from using the service.
