Thanks for trying and for sharing results here. This is a pretty interesting data point and suggests something I don't think I've read about: the possibility that safeguards may not be account-invariant; that two different users might be getting downgraded or blocked differently for the same prompts.
This raises some really interesting ethical questions in my mind, but I suppose I need to do more reading and research on this before anything else.
As someone who has worked in the space on multiple products, I can tell you with 99.999% certainty that fraud/abuse detection is never account invariant once money is involved.
Comments
I went last to first, and I was not blocked.
Thanks for trying and for sharing results here. This is a pretty interesting data point and suggests something I don't think I've read about: the possibility that safeguards may not be account-invariant; that two different users might be getting downgraded or blocked differently for the same prompts.
This raises some really interesting ethical questions in my mind, but I suppose I need to do more reading and research on this before anything else.
You bet! It's interesting.
As someone who has worked in the space on multiple products, I can tell you with 99.999% certainty that fraud/abuse detection is never account invariant once money is involved.