Alibaba reportedly bans employees from using Claude Code

In a notable move within the tech industry, Alibaba has announced that it will prohibit its employees from utilizing Anthropic’s programming tool, Claude Code, starting July 10. This decision follows reports indicating that Anthropic has already restricted access to its models for Chinese firms and foreign entities associated with them. In an effort to tighten security, Anthropic has been focusing on eliminating loopholes that previously permitted Chinese users to engage with Claude. According to insights shared on Reddit, part of this effort included a specific version of Claude Code designed to discreetly identify Chinese users. Thariq Shihipar from Anthropic elaborated in a post on X that this initiative was an experiment initiated in March aimed at curbing account misuse by unauthorized resellers while also safeguarding against a practice known as distillation, where AI models are trained on the outputs of other models. Shihipar noted that the company has since implemented more robust measures and expressed intentions to phase out this version for some time. In light of these developments, Alibaba has categorized Claude Code as high-risk software, urging its staff to transition to using the company’s proprietary tool, Qoder, instead.