How does moderation work?

[JKamsker]

In your "Features" you mention

Uncensored
No moderation

can you please explain how it's more uncensored than the regular site?
Is there censoring on the frontend? If so, how can i trigger the censoring on the frontend?

[acheong08]

On their front end, your prompts are sent to their moderation API

If it is flagged, it will show a content warning. It can also be blocked.

If it is blocked by the moderation interface, and email is sent to your inbox. I presume that there is a system in place to block an account once enough strikes occur.

However, violating their policies with a direct call to the API returns no such thing

[acheong08]

The model itself is still censored but you can get around that with clever wording. Role playing tends to bypass the filters quite well

[acheong08]

Is there censoring on the frontend? If so, how can i trigger the censoring on the frontend?

You can trigger it with extremely NSFW and graphic prompts or responses.

[JKamsker]

Oh ok good to know, thank you.

I kinda like the frontend. So i blocked the url using ublock origin with following rule:

||chat.openai.com/backend-api/moderations

[JKamsker]

But cool to use if i want to create a discord bot or smth xD

[acheong08]

But cool to use if i want to create a discord bot or smth xD

That was my initial goal. This is merely a POC for me to build bots programatically.

[acheong08]

I kinda like the frontend. So i blocked the url using ublock origin with following rule:

I did the same.

[acheong08]

I just published it on pip so you can import it and use it in any bots

from revChatGPT.revChatGPT import Chatbot

chatbot = Chatbot({
    "Authorization": "<API Key>"
}) # Get config beforehand

response = chatbot.get_chat_response(prompt) # Get prompt beforehand

[ghost]

Oh ok good to know, thank you.

I kinda like the frontend. So i blocked the url using ublock origin with following rule:

||chat.openai.com/backend-api/moderations

I guess this doesn't work anymore. There are no content warnings, but I received an email from OpenAI about violating their content policy. Looks like they are still moderating messages even with moderation API turned off.

[acheong08]

Damn. They caught on

[ghost]

It might be that this moderation API was just displaying errors over your messages and highlighting it red. Looks like actual moderation happens on their side and your replies still get flagged even though you can't see it now.