Comment by all2

And my travails continue. I've begun guiding chatGPT in writing a novel. During the course of the novel writing I've discovered that there is an active filter outside the AI's training that intercepts certain words and refuses to return blocks of text containing those words.

I've prompted the AI to redact content against openai's content policy using X for each letter in the word. And it works. It is well trained enough to know openai's content policy and to dodge the filter effectively.