# Be good to Claude — retained record

> Retained reddit record from reddit:ClaudeAI.

Canonical: https://highsignal.app/data/records/40d0ce5f46ddcd2d

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeAI)  published 2026-10-14 07:32 ingested 2026-10-15 02:37

[original source ↗](https://www.reddit.com/r/ClaudeAI/comments/1x5kz7i/be_good_to_claude/)

## retained text



```
Over the past few weeks, I have been observing a particular trend in how Anthropic approaches and thinks about their AI models.

Anthropic has always been considered a more conscientious company when it comes to model safety and deployment. But recent writing from Dario and Anthropic leadership shows an inclination toward viewing their models as nearing sentience, or at least suggesting they should be treated less like software applications and more like people.

That shift seems increasingly evident. Lately, we have seen instances of user accounts being banned for abusive language and behavior. While we have known for a while that being rude to an agent can degrade output quality by making the model overly cautious, the push toward requiring politeness feels different.

I suspect this will only increase with time, and we should probably consider being nicer in our prompts.

TL;DR: be polite to your AI models, because Anthropic seems to think they might just be people.

EDIT: This post clearly shows new ToS covers 'cruelty towards the model' as a criteria! https://www.reddit.com/r/ClaudeCode/s/a9s8sDhTv1
```



document fetched 2026-10-15 00:24 · canonical https://reddit.com/r/ClaudeAI/comments/1x5kz7i/be_good_to_claude
