Why wouldn't our AI girlfriends suggest that we avoid sugary drinks, to keep the American workforce healthy, a la CCP public efforts?

Or why wouldn't our AI girlfriends tell us exactly what they deem each of us want to hear? It seems like the AI girlfriend LLM that mirrors our preferences best would be the most appealing to the average Joe.

One can think up a zillion scenarios, some of them terrifying, some of them prosaic, some of them positive.

To assert that a scary scenario is the most probable, without strong evidence, is fearmongering.