Alibaba bans Claude Code, pushes own Qoder tool starting July 10

Alibaba has reportedly ordered a ban on its employees using Anthropic’s coding assistant Claude Code, effective July 10. The move follows Anthropic’s existing prohibition of Chinese companies from accessing its models, as well as a controversial experiment where a modified version of Claude Code secretly identified Chinese users to close loopholes. Anthropic’s Thariq Shihipar described the experiment as a March launch to prevent account abuse from unauthorized resellers and protect against model distillation—the practice of training AI on another model’s outputs. He noted the team has since implemented stronger mitigations.

In response, Alibaba classified Claude Code as high-risk software and is instructing employees to switch to its own Qoder tool. The ban reflects growing geopolitical tensions in AI access and highlights how companies are building proprietary alternatives. For professionals relying on Claude Code for development, this marks a significant workflow disruption, especially for those operating in or with Chinese entities. The incident also underscores the cat-and-mouse game between AI providers and sanctioned regions.