I was sent an email saying that my account was flagged to be in violation of Open AI policies. I was trying to summarize a report about American Indian Boarding Schools which contains documentation of sexual abuse of children and violent act committed by priests, nuns and the US Government. Such abuses have been documented by the US Dept of Interior report, “Federal Indian Boarding School Initiative Investigative Report.”
This seems to be an oversight and makes me concerned about the discernment of such policies.
Yes, OpenAI setting an AI loose and giving it the ability to ban people for policy violations and permanently remove accounts and also keep purchased credits is the exact kind of decision making that is spelled out as a violation of their own policies.
2 Likes
It is the automatic key word mapping
Reformat your article and keep a eye on key wording that might get trapped
Google does this in chat
I’m not going to start a new prepay account just to get it banned (and see that they don’t use the connection of personal information to ban other accounts).
We’ll see the trickling in of more pre-pay credit ban anecdotes you like to scorn, though.
You can search the forum or Reddit for “janitor.ai” to see the results of intelligence-based identification.
perhaps including a syntactical hint to only retrieve sources, or to rely on widely public data could help. Just speculating though- good luck!