Anthropic Exposes Large-Scale Unauthorized Claude Access Through Chinese Token Resellers

Anthropic has uncovered a sprawling operation where thousands of Chinese token resellers are providing unauthorized access to Claude, enabling local AI labs to conduct large-scale model distillation attacks. The company's September 2026 threat report documented nearly 190 million unauthorized exchanges with Claude across seven Chinese AI labs, with the reseller networks operating openly on GitHub, Taobao, and Telegram while Anthropic pursues enforcement against them.