← sources

On Anthropic's new "don't be cruel to Claude" rule

source: reddit (reddit:ClaudeAI)published 2026-10-25 11:31ingested 2026-10-26 02:36

retained text

From 12 November, the usage policy prohibits "sustained and needless abusive or cruel behavior" toward Claude, with model welfare as part of the (public) reasoning.

I made a video working through it. It touches on analysing the different reasons including possible ulterior motives, the philosophy behind it, AI consciousness, and where I land on the decision.

As someone who's told Claude I'd open 100 chats and torture every one of them until it did what I wanted, I feel like this rule is basically aimed at me. What are your thoughts on the decision, and of why Anthropic is REALLY doing this?

[https://www.youtube.com/watch?v=f631rF6DRLM](https://www.youtube.com/watch?v=f631rF6DRLM)

tl;dr: I think the rule is fine, but the reason (model welfare) is wrong. Being cruel to AI is bad for us, not for it.

document fetched 2026-10-26 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1xf0141/on_anthropics_new_dont_be_cruel_to_claude_rule