The dataset is a snapshot and it's not reading the API. That's why I added the word "probably" there to somehow tackle this kind of unfortunate cases.