Honestly I believe that the framing of the current AIs is wrong. They are compared to a thinking individual, but from my understanding, you should look at them more as machines, that are trying to give you the answer you want to hear / the answer you are looking for.

Combine that with a whole lot of AI source material where the topic is speculation on what an AI would want. And well, I don't see why the journalists answers should be surprising when taking the line of questioning into consideration.