Anthropic Adds Ban on Sustained, Needless Abuse of Claude
publishers 32articles 34first reported 8 Oct, 17:00 UTCdeveloping since 8 Oct · 9 editions
Anthropic has updated its usage policy to prohibit “sustained and needless abusive or cruel behavior” toward its Claude AI models. The new rules, announced Thursday, are scheduled to take effect on November 12.
The company said the prohibition is intended only for extreme cases in which users repeatedly behave cruelly toward its models without a discernible purpose. It does not cover ordinary frustration or pushback, dark creative themes, or model testing and research.
Claude’s ability to end abusive conversations will remain Anthropic’s primary enforcement mechanism. The company introduced that option in 2025 for rare cases of persistently harmful or abusive interactions. Anthropic’s broader enforcement rules also allow warnings, access restrictions, suspensions and account termination.
The change comes amid debate over whether advanced AI systems could be conscious or warrant protections. Anthropic has explored the issue of “model welfare,” although its updated policy does not explicitly use that term. CEO Dario Amodei has previously said the company does not know whether AI models are conscious but remains open to the possibility.
The broader policy update also consolidates restrictions on deceptive campaigns and artificial online activity, clarifies safeguards for weapons development and surveillance, and adds requirements for high-risk uses in health, finance and autonomous control of physical equipment. Anthropic also narrowed its election rules to focus on deception and disruption while removing a blanket ban on personalized voter and campaign targeting.
HOW THIS STORY WAS MADE
Written from 8 articles, 5 with the publisher's own text; 8 independent newsrooms once syndicated copies count as one; this version written 56 h after the record first saw the story; the editor kept it as written.
- Publishers
- 32
- Source articles
- 34
- Given to the writer
- 8, 5 with the article's own text
- Independent newsrooms
- 8
- This version written
- 56 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →