I don't know, maybe at worst it'll be good practice. Awkward guys can interact with someone friendly in a no-consequence way, and maybe gain some comfort.
In any case, with the exception of a few weirdos, I don't think most people are going to be confused about whether AI girlfriends are real, just as almost no one thinks a sex doll is real.
This really feels like 1650's nautical piracy. Someone outside the reach of the law of the targeted country's merchants, making tons of money by theft and ransom. And like the pirates of old, often supported by the host nation so long as their attacks disrupt the activities of rivals nations' merchants.
How uniform is the format of your dataset? Current AI tools should be pretty good at direct transformations.
If you want to generate queries, though, my guess is another 2-3 years before it's solid enough for your domains with high correctness requirements, like fintech.
It's interesting that even the current financial system fits so snugly into 2^64!
Global GDP = $100T = $10^14. If we imagine global assets being worth 10x GDP, that's $10^15.
2^64 = 1.8 * 10^19. If people care about measuring values to precision at $0.0001, that's about the limit.