Anthropic's Claude Code: China access bans and Alibaba ban

Anthropic is trying to stop Chinese companies from accessing Claude Code, enforcing terms of service that explicitly ban sales to companies controlled by China. At the same time, Alibaba has banned its own employees from using Claude Code and required them to delete all Claude models, after reports of hidden code that could flag China-linked users.

What restrictions has Anthropic placed on Claude Code?

Anthropic's terms of service explicitly ban sales to companies controlled by China, and the company has moved to prevent Chinese firms from accessing Claude Code. The company has also accused Alibaba, DeepSeek, Moonshot AI, and MiniMax of using Claude for "distillation," meaning training smaller models on Claude's outputs, which Anthropic says it is trying to stop.

Anthropic is taking both contractual and enforcement steps. The terms bar sales to China-controlled companies; Anthropic has alleged specific instances of distillation by named firms as part of the evidence it cites to tighten controls.

How are Chinese companies getting around the restrictions?

Chinese companies are reportedly bypassing Anthropic's restrictions through cloud services, overseas subsidiaries in Singapore, or VPNs. Examples named in the article include Ant Financial and ByteDance using those workarounds to access Claude Code despite the sales ban to China-controlled entities.

Those technical and corporate routes undermine a straight contract-based block. Cloud providers, foreign-registered subsidiaries and virtual private networks can be used to route traffic or host services outside mainland China, allowing firms to reach Claude Code in ways Anthropic's raw terms do not automatically block.

Why did Alibaba ban Claude Code internally?

Alibaba banned its employees from using Claude Code and required deletion of all Claude models after reports surfaced of hidden code in Claude Code that could flag users based in China or linked to a Chinese lab. Anthropic's Thariq Shihipar described the flagged-code incident as "an experiment from March to stop account abuse and distillation," and said that stronger safeguards have since replaced it.

Alibaba's internal prohibition came amid those reports and follows Anthropic's prior accusations of distillation by multiple companies. The combination of alleged hidden detection logic and the distillation dispute appears to have prompted Alibaba to take a preventive internal step.

Why it matters

This episode exposes a growing enforcement gap between contractual restrictions and technical workarounds. Anthropic's ban on sales to companies controlled by China is precise on paper, but cloud services, overseas subsidiaries and VPNs provide practical methods to bypass those restrictions. At the same time, reports of hidden code that could flag China-linked users raise corporate trust and compliance questions that employers such as Alibaba find serious enough to enact internal bans.

The clash also shows how disputes over model distillation are escalating into corporate and geopolitical disputes: Anthropic has named Alibaba, DeepSeek, Moonshot AI, and MiniMax in accusations that tie intellectual property enforcement to national controls on AI access.

What to watch

Watch for any public responses or policy updates from Anthropic that go beyond terms of service enforcement, and for whether Alibaba or other named firms issue fuller explanations or reverse their internal restrictions. The next concrete signals will be formal takedown notices, technical blocks by cloud providers, or new safeguards Anthropic publishes to prevent distillation and account abuse.