Gray-market resellers in China sold Claude access at a 90% discount

In China, a gray market of proxy services sold access to Claude, Anthropic's flagship model, at a 90% discount off the official price. The finding comes from a study by Oxford China Policy Lab researcher Zilan Qian.

How the scheme works

The traders used several revenue techniques at the same time. They ran proxy services on stolen Anthropic accounts, giving users access to Claude at a 90% discount. But that was only the start: they then swapped in cheaper models (GPT-3.5 or local ones) and pocketed the difference. Most importantly, they resold user queries and responses to third parties for distilling their own AI models.

• Use of stolen Anthropic accounts
• Substituting cheaper models for Claude
• Collecting and reselling user data
• Official price + 90% discount = profit on every customer

Why this is a problem

For users, this means their queries end up in uncontrolled hands. Their data is used to train competing models. For Anthropic, it means a loss of control over its model in one of the largest AI markets, along with gross losses. The study showed the scheme was active and found demand: users liked the low price, and sellers liked the profit. Chinese internet law enforcement usually does not pursue such schemes aggressively when they involve foreign companies.

What this means

It shows a classic conflict between market openness and IP control. Anthropic cannot effectively protect Claude in places where the state has no interest in protecting it. Chinese companies benefit from harvesting data for their own models. The global AI economy is still a Wild West without common rules.