@sindarina I'm not going to claim it's "AI", but I do think it's probably not what I would call as [single] "photograph" - that is to say, a capture of the entire image from a single exposure over the same interval of time. The dynamic ranges of the different elements don't add up.
It's likely a HDR-style composition from what I would call "multiple photos" and in general (not particularly in the setting here) this kind of technique tells counterfactual stories as if they were fact.