Breaking

Chinese Claude Reseller Market Draws Scrutiny as Anthropic Targets Unauthorized Distillation

October 5, 2026 at 09:30 EDT

Foundation Models
AI Agents
Security

A gray market of Chinese token resellers is keeping Anthropic's Claude accessible to AI developers despite the company's restrictions, raising fresh concerns that frontier U.S. models are being used to train competing systems without authorization. New reporting says thousands of intermediaries are selling access to Claude and other U.S. models, while Anthropic's enforcement actions are pushing some labs toward shell companies and other workarounds.

October 2026 · Anthropic

Claude's Gray-Market Access Problem

Chinese token resellers and proxy networks are keeping Claude reachable for restricted users, while enforcement pressure is pushing some activity toward shell companies and harder-to-trace access routes.

The reported scale has expanded

February campaigns

16M+

Claude exchanges

Alibaba-linked campaign

28.8M+

Claude exchanges

The taller column is proportional to the larger reported exchange count.

24,000+

fraudulent accounts described in the February cases

25,000

almost this many accounts in the Alibaba-linked allegation

10%

reported low-end resale price versus official access

How access moves

Reseller

Pools or buys account access.

→

Proxy layer

Routes requests and may log data.

→

Model lab

Uses outputs for capability extraction.

Why users try it

Developers get cheaper access to frontier-model capabilities where direct commercial service is restricted.

What can go wrong

Requests can be substituted, credentials can be exposed, and prompts may be harvested for training data.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.