Here's my real point: this is a girl the guy loves, and I assume trusts. Maybe that's not true, but I think it's reasonable to assume, or else why would he so easily give up his passwords? Well, you've just questioned her faithfulness and honesty. Maybe she'd dump him, but to suggest she might get pregnant with another man, have the other man's child, and have him raise it? That is more sinister than even the average simpleton's life. I would agree with your general point when looking at the whole of humanity, because many people aren't honest, do cheat, and do follow their most base desires.
I like to think the type of people visiting bay12games have a good eye for decent people, and thus girlfriends. Might not be true, but I think it's preferable to assuming the worst. If neither of the two were reading the topic at all, I wouldn't consider the implication disrespectful. Consider this; his girlfriend may be reading your lines and has now just had you question her honesty and faithfulness by the mere fact she is a woman. You don't know anything about her, her principles, her morals, her care for him, yet still felt compelled to question her honesty and faithfulness, putting them up in the air. The "going to your family" was meant that you may have just indirectly told his family (his girlfriend) that you have enough doubt in her faithfulness to warrant raising. Maybe a poor analogy, I was merely stating you probably wouldn't want strangers planting seeds of doubt in your family's trust of you.
One last thing, I'd like to object to the idea attractiveness of a man or woman is a definitive indicator of their health or "better" genetics. After all, most of us would consider someone highly intelligent to have good genes, but as far as evolution is concerned, all that matters is the amount of successful offspring you pass on. Someone could live longer, be healthier in middle age, be more intelligent, and yet be less fertile. Who really has the worst genes, when it comes to a conscious evaluation? Our preferences are because of correlations (X feature has Y probability of Z), and that means they can deceive us when it comes to what we would commonly measure as "good genes".