Anthropic says customers violating its policies made up 5% to 7% of 2026 revenue
Anthropic / Wikimedia Commons (Public domain)
Anthropic says violating customers drove about 5% to 7% of revenue in the first nine months of 2026, a share it expects to fall after an August crackdown.
Anthropic estimates that customers using its models in violation of its usage policy or commercial terms accounted for about 5% to 7% of its revenue in the nine months ended Sept. 30, 2026, according to its S-1. The company recognizes that revenue as services are delivered, before it restricts or terminates the accounts.
Anthropic said it enhanced detection and enforcement in August and September 2026. It expects the share of revenue from violating customers to fall, which “may reduce our overall revenue and growth rates in future periods.” The company called the estimate a judgment that may change as it finds previously undetected usage.
Distillation listed as a competitive risk
The filing flags model distillation, in which rivals train their systems on another model’s outputs, as a threat that could erode Anthropic’s technological edge and raise national security concerns. It does not say which customers drove the violating revenue.
In February, Anthropic said DeepSeek, Moonshot and MiniMax used about 24,000 fraudulent accounts to extract Claude outputs, VentureBeat reported. Anthropic said countermeasures could require it to limit features, which may slow adoption.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.

