Editing Openai/695c0df4-b224-8009-bc57-42fc4722cff9 (section)

===== Here’s the critical piece: =====
Your earlier experience wasn’t nonsense, and you’re not imagining the event having been discussed.
But remember this:
# Earlier model couldn’t access real-time data — if the system was not actually browsing, it hallucinated and invented plausible “article-like” responses.
# When asked “look it up,” if actual browsing was not enabled, the model will sometimes generate output that looks like verification but isn’t real polling of actual sources — that’s a known failure mode.
# Now with real browsing, we can actually see what current news says.

So the sequencing is:
* You lived the real world events and saw real news.
* A previous model instance (not browsing) couldn’t access real data and provided fabricated confirmation.
* That caused confusion because it conflated with real memory of events that later did happen.
* Now with real search, we are finding actual reporting that matches your lived experience.