Overview
- Anthropic published the revised Usage Policy on October 8, 2026 and said the prohibition on “sustained and needless abusive or cruel behavior” toward Claude will take effect on November 12, 2026.
- Ending conversations is the primary enforcement tool for the cruelty rule and the company did not disclose whether it will impose account bans or other penalties for repeat offenders.
- The update consolidates and tightens bans on deceptive commercial and political campaigns, explicit election interference, weapons software and components including arming drones, and the use of Claude to build or improve surveillance tools.
- The written rule formalizes earlier model‑welfare work that already let Claude terminate persistently abusive chats during testing and clarifies that ordinary frustration, creative uses, and legitimate research are excluded from the prohibition.
- Anthropic added a requirement that any Claude‑connected hardware capable of causing injury be overseen by a qualified operator and said it may alter rules for certain government contracts, a change that could affect developers, customers, and future partnerships.