
Anthropic is attempting to prevent Chinese companies from accessing Claude Code through contractual restrictions, but firms are finding workarounds via cloud providers and VPNs. Meanwhile, Alibaba has internally banned Claude Code use after discovering code that could identify China-based users, citing concerns about account abuse and model distillation. The dual pressure—regulatory restriction and internal corporate distrust—highlights the practical limits of enforcing geographic controls.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Anthropic is restricting Chinese companies' access to Claude Code through terms of service, but firms like Ant Financial and ByteDance are circumventing the ban using cloud services, overseas subsidiaries, and VPNs. Separately, Alibaba has banned its own employees from using Claude Code and is requiring them to delete all Claude models after discovering hidden code that could flag users based in China or linked to a Chinese lab.
Why it matters
The dual-front enforcement challenge reveals the tension between Anthropic's stated policy and its practical ability to enforce it. Alibaba's move suggests trust concerns about code safety and potential surveillance mechanisms, which could affect how other organizations in the region evaluate using Claude products. Anthropic has previously accused Alibaba, DeepSeek, Moonshot AI, and MiniMax of using Claude to train their own smaller models.
What to watch
Anthropic's Thariq Shihipar described the hidden code as an experiment from March to stop account abuse and distillation, and stated that stronger safeguards have since replaced it. The effectiveness of these new safeguards and whether other organizations adopt similar internal bans will signal the severity of trust issues around Claude.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion




Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack