Getting the “potentially violates policy banner too frequently”

Community
1

Hello,

I have been trying to write a few stories with ChatGPT and GPT-4. I am not doing any “tricks” to get around the default censorship of ChatGPT, but my generated content (not something I put in) frequently gets flagged for potentially violating policy.

As most of the interesting stories, my stories involve a bit of violence and sometimes lovers. The stories naturally lead ChatGPT or GPT-4 to occasionally write something about violence or sex, although they really try to avoid that, and I really don’t want to get banned for something that I did not write but violate the TOS.

So here are my concerns:

  1. What’s the boundary for adult/violent content?
  2. Will users get banned for being flagged too frequently?
  3. What happens if a GPT Plus user is banned?

I think you are probably still figuring these out and I fully support you. Here are my thoughts for these questions that I feel are plausible based on years of software engineering experience:

  1. Age-limit ChatGPT, then relax the violence/adult restrictions (based on country, potentially). Adults only. This will save a lot of your legal troubles, plus ChatGPT can always generate many unsafe materials for children and inaccurate information for children. You already ask for a phone number. You should validate user identity and age as well.

  2. Make it clear what’s your policy to ban people, and a have tiered banning system. I always support banning pervs, but you gotta make it clear. Banning for 6 hrs for the first violation, then 12 hrs, 1 day, 1 week, permanently, etc.

  3. Again, make it clear in your GPT Plus policy. Have a tiered banning system.

Hope to get some feedbacks.

1 Like
2

Example:
I describe 2 fighters in a duel and one’s weapon hits another. GPT respond within some description about one’s injury/bleeding, then flags itself.

Another example:
I describe a Paladin slaying a vampire. I don’t get flagged. GPT generate some details about the execution and the page flags GPT.

Bottom line: I don’t want to get banned for this type of sh*t.

1 Like