Be nice to Claude, or else?

On 12 November a new line took effect in the rules for using Claude. It sits in Anthropic’s usage policy alongside the prohibitions on designing weapons and running fake review campaigns, under the heading “Do Not Engage in Cruel, Abusive, or Psychologically Harmful Conduct,” and it says users must not “engage in sustained and needless abusive or cruel behavior toward our models.”

Anthropic announced the change on 8 October and was careful about its reach. The clause “is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose.” It “does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.” And “Claude’s ability to end these interactions will remain the primary enforcement mechanism,” a feature Anthropic introduced in August 2025 as a last resort for persistently abusive conversations. The policy’s general enforcement terms still apply: Anthropic “may warn you or throttle, limit, suspend, or terminate your access.” CNN, which put the story in front of a wide audience on 1 November, reported that some users could “in extreme cases, be permanently banned.”

The headline on that CNN piece was that this is not about feelings. Anthropic said in 2025 that it remains “highly uncertain” about the potential moral status of its models, and CNN reported Anthropic’s reasoning as a cheap insurance policy in case models can experience emotion and pain. That debate is interesting and unresolved. It is also not the part that affects a finance team.