The article is from August, 2023. Of course AI's can do it now; they are trained on lots of articles written about the spotless baby giraffe. The point is whether or not they were able to do it before humans had analyzed it.
You think that the reason it can now is because of some articles that were written about it. From my tests it looks like it can understand the image, its contents, and what's missing. I even asked if it knew about any articles about it (without using web search) and, no, it doesn't. Why do you make baseless assumptions like that? Just try it out instead of speculating.
The model is not capable of identifying whether it knows a fact or not.
I think the deeper issue is that we didn't know a priori whether these systems could handle this prompt. One way I've heard this described is that the frontier of what computers can do has advanced, but the exact boundary is difficult to know. I don't think one anecdote really makes it clear.