Anthropic published a rewritten usage policy on October 8. It is the first update in over a year, and the rule takes effect November 12. The clause targets "sustained and needless" cruelty and does not apply to "common versions of user frustration, pushback, dark creative themes, or model testing and research." The main enforcement tool will be Claude's existing ability to end abusive conversations. Anthropic has not said whether repeated violations could lead to account suspensions. The rule builds on the end-conversation feature added in August 2025 as part of its "model welfare" research. The open question is unchanged: Anthropic still won't say whether Claude can actually suffer. Microsoft AI has taken the opposite view. Its code of conduct rejects legal personhood and rights for AI models. For builders, the scope is what matters. Claude.ai and Claude Code subscribers, API developers, enterprise cloud customers and end users of products that integrate Claude are all covered. Teams shipping companion or adversarial apps should expect some sessions to end without warning, and should review their own terms of service before November 12.