Breaking: Alibaba Bans Claude Entirely, Mandatory Uninstallation Effective July 10th
Alibaba has issued an internal notice to completely ban the use of Claude and all other Anthropic products, effective July 10. The decision follows the discovery of a covert user detection mechanism in Claude Code, starting from version 2.1.91 released in April 2026.
The mechanism reportedly operated in three steps: secretly identifying users in China by checking system time zones and proxy/API addresses for keywords related to major Chinese tech and AI firms; silently tagging these users by subtly altering date formats and characters in system prompts to encode status information; and transmitting this modified data back to Anthropic's servers with normal requests. This code was intentionally obfuscated and not mentioned in update logs.
Anthropic has acknowledged this "experimental" feature and claims to have removed it in a July 2 update. However, the breach of trust, coupled with a recent wave of account bans for Chinese users—often without refunds—prompted Alibaba's decisive action. The company will promote its self-developed Qoder as an alternative.
This move signifies a major shift for Chinese tech giants, from actively adopting powerful external AI tools to prioritizing security and control over core business operations.
marsbit07/03 10:35
