Alibaba Bans Employees From Using Anthropic's Claude Code

Alibaba has reportedly issued a directive banning its employees from using Anthropic’s powerful programming tool, Claude Code, effective July 10. This move marks a significant escalation in the growing tension between Western AI providers and Chinese tech giants over access, security, and intellectual property.

A Shift Toward Internal AI Ecosystems

In response to the restrictions, Alibaba has officially classified Claude Code as "high-risk software." Rather than leaving developers without advanced coding assistance, the company is mandating the use of its proprietary AI tool, Qoder. This strategic pivot underscores a broader trend in the Chinese tech sector: the aggressive push toward self-reliance in Large Language Models (LLMs) to mitigate the risks of being cut off from Western-developed frontier models.

By forcing a migration to Qoder, Alibaba is not only attempting to secure its internal workflows but is also reinforcing its own AI ecosystem, reducing dependence on external providers that may be subject to geopolitical volatility.

Closing the Loopholes: Anthropic’s Defense Strategy

The ban comes amid intensifying efforts by Anthropic to enforce its existing policies, which prohibit Chinese companies—and foreign entities owned by them—from accessing its models. Anthropic has been actively working to close technical loopholes that previously allowed users in restricted regions to bypass access controls.

Technical details suggest these efforts included an experimental feature within Claude Code designed to identify Chinese users. According to Anthropic’s Thariq Shihipar, this mechanism was part of a March experiment aimed at preventing account abuse by unauthorized resellers and protecting against "distillation." In the context of AI, distillation is a sensitive practice where one model is trained using the outputs of another, a process that can lead to intellectual property theft or the unauthorized replication of a model's capabilities.

Why This Matters for the Global AI Landscape

This development is a microcosm of the "AI Cold War" currently shaping the global technology industry. As companies like Anthropic implement stronger mitigations to prevent model distillation and unauthorized access, the boundary between globalized tech development and regionalized "AI islands" becomes more pronounced.

For developers and founders, this highlights a growing operational risk: the tools used to build software today may become unavailable tomorrow due to regulatory shifts or corporate compliance policies. As major players like Alibaba move toward internal alternatives like Qoder, the industry may see a permanent fragmentation of AI tooling, where developers must choose between the cutting-edge capabilities of Western models and the localized security of regional ecosystems.

Key Takeaways

  • Mandatory Tool Migration: Alibaba is classifying Claude Code as high-risk and directing all developers to use its in-house tool, Qoder, starting July 10.
  • Mitigating Model Distillation: Anthropic is tightening access controls to prevent "distillation," where companies use AI outputs to train competing models.
  • Increasing Fragmentation: The move signals a deeper divide in the AI industry, as major tech firms prioritize self-reliance and proprietary ecosystems over globalized AI services.