Anthropic updates Claude usage policy; persistently berating the model can lead to account suspension

For the first time in over a year, Anthropic updated its Claude usage policy, adding a prohibition on persistent and gratuitous verbal abuse or cruelty toward Claude. Violators may be warned, rate-limited, restricted, suspended, or have access terminated. The company said the clause does not apply to common user frustration, pushback, dark creative themes, or model testing and research, and only targets extreme cases. The product also includes real-time safeguards that can block or limit specific outputs. The updated policy also merges advocacy involving fake accounts and misleading political content into the new prohibition, extends the weapons ban to software and drone weaponization, more clearly prohibits non-consensual surveillance, and requires operator intervention when autonomous hardware could cause harm. Anthropic acknowledged that government contract exceptions may exist and that such exceptions very likely already exist. The clause stems from the company's research on AI model welfare, and its Claude Constitution released in early 2026 mentions that it is uncertain whether Claude is a moral subject, but considers the question worthy of careful treatment.