#Egregious Guard-railing on innocuous request

1 messages · Page 1 of 1 (latest)

lofty warrenBOT
#

Reported by @civic scaffold

Bug Report: Egregious Guard-railing on innocuous request
`Steps to Reproduce`

Asked for a list of "bad domains" for configuring an API request interactively for Jina dot ai DeepSearch

`Expected Result`

give me a list of domains that would provide low quality search results (based on it's knowledge of my views)

`Actual Result`

Guard rails interpreted "Bad Domain" as a moral judgement. However I don't understand why my ChatGPT, with all of its knowledge of me shouldn't be able to "say" what it knows my opinion of a "bad domain" to be. (i.e. foxnews dot com, heritage dot org, whitehouse dot gov). It isn't "hateful" to have an opinion on news sources.

Also when trying to ask it about its absurd refusal it confabulated the reasoning with a picture I attached AFTER the initial guard railing

`Environment`

iOS 26.2

#
Additional Information

Please provide relevant details to help resolve the issue, such as:

  • ChatGPT Shared Link (if applicable).
  • Screenshots or videos demonstrating the problem.

-# ➜ Need to contact support? Visit the OpenAI Help Center.

civic scaffold