New Delhi, October 5, 2026 (Yes Punjab News)
An underground market providing access to Anthropic’s Claude AI model has expanded rapidly in China, where the service is officially blocked, with brokers reportedly reselling accounts to domestic users, according to a report.
South Korean publication The Chosun reported that the trade is driven by “token resellers” who establish servers in countries where Claude is available, accumulate large numbers of accounts and then sell access to Chinese users.
The buyers reportedly include scholars and engineers who favour Claude for code generation and want early access to newer versions of the model.
The report, citing US technology publication The Information, said some resellers use large numbers of email addresses obtained through earlier cryptocurrency ventures to register Claude accounts.
The scale of the trade is also reported to be significant. As many as six of 30 companies in an office building in Beijing’s Haidian District are allegedly involved in AI access reselling, while tens of thousands of such entities are said to operate across China.
With US-based AI platforms tightening access controls, resellers have reportedly adopted more elaborate methods to bypass restrictions. These include establishing shell companies in Southeast Asia, the Middle East and the United States to enter enterprise agreements with Anthropic or Amazon Web Services.
“These processes occur outside China, making tracking difficult. It’s a game of hide-and-seek between those blocking accounts and those misusing them,” the report said.
The illegally obtained accounts are also reportedly being used by Chinese companies for what is described as “distillation learning”, a process in which the outputs of an advanced model can be used to help develop another AI system.
In March 2026, Anthropic accused three Chinese AI companies, including DeepSeek, of improperly extracting capabilities from Claude to advance their own models. According to Anthropic, the alleged operation involved about 24,000 fraudulent accounts and more than 16 million exchanges with Claude.
Anthropic has warned that models developed through such methods could lack the safety safeguards built into systems such as Claude, potentially increasing risks of misuse, including for cyberattacks and biological weapons.
FOLLOW NOW
