Anthropic’s S-1 filing disclosed that a portion of the revenue the company earned during the first three quarters of 2026 came from accounts that breached its commercial terms. The artificial intelligence startup estimated that customers who violated its usage policies or commercial terms made up 5% to 7% of its revenue in that period, though it expects that revenue to cease as access for those customers is restricted or terminated.
The company did not specify the nature of the violations in the filing, including whether the income was linked to customers based in China or to the practice of adversarial distillation, a method used to create competing models.