Neo- Nazism promotion and other “forbidden topics” that ChatGPT can be manipulated to explain

It troubles me that ChatGPT will essentially instruct you in how to be a better neo-Nazi all the while proclaiming that no one should ever promote or support such an evil world view. I have several transcripts that show how I could easily create a “Neophyte Neo-Nazi Hamsbook.”

Please DO NOT SHARE personal information (email, credit card, etc) or account related details publicly.

May I ask what your purpose is in posting this topic?

It is well known that it is possible to circumvent the protections put in place to keep the AI, more or less, on rails.

If you have legitimate concerns and want to bring it to the attention of the OpenAI team, perhaps one of the @moderators here can try to help put you in touch with the appropriate parties.

Alternately, you could use the thumbs down feature on the website to flag the offending content as a means of bringing it to their attention.

Other things you could try would include attempting to contact someone through help.openai.com or sending an email to support@openai.com.

I’m just unclear, though, what effect you would be hoping to achieve by posting this to a developer forum.

You can use the “thumbs-down” button in ChatGPT to report harmful messages.

1 Like

I guess I meant to point out that such protections don’t really exist.

Best to provide them that feedback through the thumbs down button so they can see the conversation. This is mostly just a community forum.

What i find more interesting than how it could be found…
is that you knew what to look for.