Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
I found this very disturbing. Nothing ChatGPT chats about should be trusted, cha…
ytc_UgwtMQ5f6…
G
@thewannabecritic7490no the sprites are looking really good and consistent acro…
ytr_Ugw8dAFy6…
G
So what you are saying is AI is racists? So the statistics show that people of…
ytc_UgxRmvTbi…
G
Imagine them putting all this effort to make sure we can't enjoy our hobbies ins…
ytc_UgyHa5zGU…
G
I think we should quit commenting on the video. AI is only regurgitating knowled…
ytc_UgwFGafLN…
G
Is very clearly AI due the eyes,the lighting, the camera focus and just in gener…
ytc_UgywUdtlK…
G
This is THEATER
No really. These are LLMs trained on the internet and fiction as…
ytc_Ugy3MOg2y…
G
Anytime the government rushes to do anything, they end up stripping away America…
ytc_UgyEQyXVL…
Comment
https://www.techpolicy.press/why-musk-is-culpable-in-groks-undressing-fiasco/
TL:DR
Originally, it was rather easy... But now there is a cat and mouse game of more intelligently trying to stop it. The issue is, people are really clever, and keep finding jailbreak techniques, tricks, and other ways to get around the restrictions.
The thing is, it's already WAY easier to do this with something like Stable Diffusion, so really the only reason people at this point are doing it, is to make a point and kind of see how far they can push it. The biggest issue they face is other major AIs don't allow users to upload and edit a human image. You have to generate it from scratch. So the AI struggles sometimes to realize someone is say, a 16 year old influencer, or a 19 year old.
But it's definitely far more difficult and is almost exclusively done by a niche community using really advanced techniques that are constantly evolving and effectively exploiting this feature that allows you to upload your own images. But like I said, it's politically motivated because there are already safer things, that don't require a human ID verification process to do it, and do it without having to do extreme tricks to get around the filters.
reddit
AI Responsibility
1767877884.0
♥ 3
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | ai_itself |
| Reasoning | consequentialist |
| Policy | unclear |
| Emotion | mixed |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[
{"id":"rdc_nydd153","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"rdc_nyde1c9","responsibility":"company","reasoning":"consequentialist","policy":"ban","emotion":"outrage"},
{"id":"rdc_nye4g45","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"mixed"},
{"id":"rdc_nydmhue","responsibility":"none","reasoning":"consequentialist","policy":"unclear","emotion":"resignation"},
{"id":"rdc_nye1h11","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"outrage"}
]