Alibaba orders company-wide Claude uninstall by July 10
TL;DR
Alibaba told all staff to uninstall Claude Sonnet/Opus/Fable and Claude Code by July 10, after Anthropic accused it of running 28.8M distillation exchanges through 24,800 fake accounts.
Alibaba issued an internal notice on July 3 banning all Anthropic products company-wide, ordering staff to uninstall Claude Sonnet, Claude Opus, Claude Fable and the Claude Code agent by July 10. The ban covers internal code environments and personal machines; violations are treated as compliance incidents.
The policy is a full reversal from six months ago. Alibaba spent early 2026 pushing AI adoption by reimbursing employees for external model spend — Claude, GPT-5 and Gemini were all on the menu, and weekly API bills of a few hundred dollars per engineer were common. Claude Code was widely embedded across Alibaba Cloud and Taotian engineering teams; parts of the core code-review pipeline ran on Claude Opus.
The trigger is a June 10 letter Anthropic sent to the US Senate Committee on Banking, Housing and Urban Affairs. Anthropic accused Alibaba of running roughly 24,800 fake accounts against Claude between April 22 and June 5, generating over 28.8 million exchanges aimed at extracting software-engineering and agentic-reasoning capabilities. Anthropic called it «the largest known distillation attack on Anthropic to date». The same letter had already named DeepSeek, Moonshot and MiniMax in February; Alibaba is by far the biggest name on the list.
Alibaba has not publicly responded to the accusation. The ban lands on another timeline: Alibaba sued the US Department of Defense in June to be removed from the Chinese Military Companies list. The «experimental» access restrictions Anthropic put on Chinese users starting March 2026 had been on Alibaba's internal compliance watchlist for months.
After July 10, Alibaba employees are left with GPT and Gemini on the external side. The internal notice did not say whether Qwen will pick up the budget Claude Code is vacating.
via BlockBeats / IT Home / CNBC / Tom's Hardware
The policy is a full reversal from six months ago. Alibaba spent early 2026 pushing AI adoption by reimbursing employees for external model spend — Claude, GPT-5 and Gemini were all on the menu, and weekly API bills of a few hundred dollars per engineer were common. Claude Code was widely embedded across Alibaba Cloud and Taotian engineering teams; parts of the core code-review pipeline ran on Claude Opus.
The trigger is a June 10 letter Anthropic sent to the US Senate Committee on Banking, Housing and Urban Affairs. Anthropic accused Alibaba of running roughly 24,800 fake accounts against Claude between April 22 and June 5, generating over 28.8 million exchanges aimed at extracting software-engineering and agentic-reasoning capabilities. Anthropic called it «the largest known distillation attack on Anthropic to date». The same letter had already named DeepSeek, Moonshot and MiniMax in February; Alibaba is by far the biggest name on the list.
Alibaba has not publicly responded to the accusation. The ban lands on another timeline: Alibaba sued the US Department of Defense in June to be removed from the Chinese Military Companies list. The «experimental» access restrictions Anthropic put on Chinese users starting March 2026 had been on Alibaba's internal compliance watchlist for months.
After July 10, Alibaba employees are left with GPT and Gemini on the external side. The internal notice did not say whether Qwen will pick up the budget Claude Code is vacating.
via BlockBeats / IT Home / CNBC / Tom's Hardware
