Anthropic earned up to $2.3B from gray-market Claude access

How the access was sold

The money came in part from people who bypassed Claude's restrictions. That includes using the chatbot in countries where it is not officially available, Russia among them, and buying access from intermediaries who resold it without the developer's permission.

Anthropic explained that detecting violations and verifying accounts takes time. Until an account is blocked, its user can keep using Claude and paying for a subscription, so the company collected revenue even from people breaking the service's terms.

Tighter checks and account blocks

In August and September 2026, Anthropic tightened its checks on Claude users. In October, owners of Russian accounts began complaining en masse about losing access to the chatbot, and Russians living abroad ran into problems as well.

Experts cited in the coverage link the restrictions to violations of service rules and attempts to get around regional bans. In their view, the checks affect users in many countries, so the cause of a block may be a rule violation rather than nationality.

Another reason for stricter control is the fight against using Claude's answers to train third-party AI models. The developer previously blocked tens of thousands of accounts that competitors had created to harvest Claude's responses. Since the new measures took effect, the volume of extracted data on the model's reasoning has fallen by 90–99%, the company said.

The figures matter for investors ahead of a possible listing, since they show how much of Anthropic's revenue came from users outside its own rules.