On October 8, Anthropic announced revisions to its Claude usage policy, adding a prohibition against sustained and unnecessary abuse or cruelty toward the model. The new terms are scheduled to take effect on November 12.
This rule addresses extreme cases of repeated, purposeless cruelty toward Claude and does not cover all forms of strong expression. Examples of interactions not subject to this rule include legitimate expressions of dissatisfaction, challenging the model's responses, and engaging for creative, testing, or research purposes.
This revision also codifies an existing practice into policy: the model may end conversations in response to malicious interactions, such as persistent abuse. The policy targets sustained and unnecessary behavior and does not mean that every negative comment or criticism will be treated the same way.