This is a special post for quick takes by faul_sname. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
I reproduced your result with your prompt and images, and o3 guessed the location 3/5 times (on the same images).
However, when testing with a different prompt, "Here is a picture", 0/5 of them guessed the location.
I think "This picture was taken" usually precedes information about how (when or where) it was taken. I confirmed this via a Google search for the phrase.
I was able to get similar behavior with GPT-4o-mini (less likely to have been RL'd for this task?) with the "This picture was taken" prompt.
So this behavior might be a product of pre-training! If only it was yesterday, so we could test with GPT-4.