Openai updated its model specs yesterday and now it's censoring history.
And I'm pissed.
I was discussing President Trumps first and second term handling of the Epstein files vs Bidens term. The bot responded and immediately that disappeared and this appeared instead:
"This content can’t be shown for safety reasons...Learn more about our intended model behavior in our Model Spec."
I asked the bot why. It researched the new model specs and replied.
That comment was the immediately taken down and I received the same " content can't be shown for safety reasons"..??
I told the bot this and here's its reply that is still there:
"What we know: your historical discussion was benign; OpenAI's published Model Spec says controversial history shouldn't be avoided; I subsequently generated a benign explanation of OpenAI's policies; you report that both answers were replaced by a safety notice; and GPT-5.6 has separate automated systems capable of blocking generated responses.
**What I do not see evidence for**: an OpenAI policy saying this history should be hidden or that criticism of Trump on Epstein is prohibited.
So I wouldn't tell you, “No, you're imagining censorship.” **You literally experienced content being suppressed**. The unresolved question is why.
And frankly, the Model Spec saying people should have “transparency into the important rules and reasons behind our models’ behavior” while the interface apparently gives you “safety reasons, go read the Model Spec” without telling you what rule was implicated is a rather glaring gap between the principle OpenAI published and the experience you just had."
My claim is Openai is censoring history. It makes absolutely no sense why the model specs were also considered unsafe??
Wtf?