This is just my take after trying too much to make it work.
Background:
Too many times that with so many users, I just lost count with how this “Gpt” flagged something and refused to work, or work sub-optimally due to their heavily sanctioned “wordlists” they deem “offensive”.
This creates a downward spiral where you can encounter unexpected errors where it should be none if it’s just humans. I examined that gpt also has dual layer protections, namely inputs and outputs, so anything that they deem “offensive” and will be flagged so, making works take longer and convoluted.