Story perspectives

Anthropic Accuses Chinese Labs of Covertly Routing Claude Requests

By Drooid · Sep 11, 2026, 1:31 PM UTC · How we work

Summary

Story summary

Anthropic says Chinese AI labs covertly routed user requests to Claude models.
Alibaba generated over 151 million Claude exchanges between May and July 2026.
Alibaba peaked at nearly 3 million daily exchanges from over 3,500 fraudulent accounts.
Moonshot sent about 300,000 requests to Claude in ten days using 5,380 fraudulent accounts in Singapore and Japan.
Anthropic recorded more than 12 million distillation attacks linked to DeepSeek in July 2026.

History

Historical Context

On March 23 2016 Microsoft launched Tay, a Twitter chatbot that learned from user input. Within 16 hours and after posting more than 96,000 tweets, coordinated trolls forced its shutdown, exposing how unchecked language models can be weaponized. The incident prompted regulators to demand tighter controls, shaping today’s global oversight of AI chatbots.

1 of 2

Story summary

Anthropic says Chinese AI labs covertly routed user requests to Claude models.
Alibaba generated over 151 million Claude exchanges between May and July 2026.
Alibaba peaked at nearly 3 million daily exchanges from over 3,500 fraudulent accounts.
Moonshot sent about 300,000 requests to Claude in ten days using 5,380 fraudulent accounts in Singapore and Japan.
Anthropic recorded more than 12 million distillation attacks linked to DeepSeek in July 2026.

1 / 2