# Since when does ChatGPT makes up own inside jokes? — retained record

> Retained reddit record from reddit:ChatGPT.

Canonical: https://highsignal.app/data/records/34ef2fe611f929c3

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:ChatGPT)  published 2026-09-03 17:45 ingested 2026-09-04 02:38

[original source ↗](https://www.reddit.com/r/ChatGPT/comments/1w6eh1b/since_when_does_chatgpt_makes_up_own_inside_jokes/)

## retained text



```
Hi,

some background to understand why its behavior kind of spooked me out.

So i am using chatgpt since two years roughly, and the custom instruction says, no emojis, emoticons, a seriously, professional tone, no jokes etc. 

Which works really good. In the two years no emojis or pseudo funny texts. Only clean, smooth, straight-to-the-point answers.

Now several thing came together, i use headroom for the codex cli, and i have regular problems with headroom and i use chat to solve these problems (because i dont want to waste tokens on codex for this).

Now, today i was discussing a scenario with Chat, the topic was how to find specific talents for very specific positions and it answered literally:

 "I wouldnt simply use Headroom... er, Headhunter 😉 here." 

...like, what? That is so out of ...Character...if something like this exist in the context of ai. 

It never did a joke once before and didnt do it again after this conversation. That caught me so off guard, that i dont know what to think of it, i mean why did it do that? It is a software program which follows rules and it didnt follow them.

Dont know what to hope for after writing all this. I am just a little bit distraught for some reason and wanted it to get it off my chest lol
```



document fetched 2026-09-04 00:25 · canonical https://reddit.com/r/ChatGPT/comments/1w6eh1b/since_when_does_chatgpt_makes_up_own_inside_jokes
