The Dangers of Using AI for International Planning
Contents
Someone wrote us a few days ago with an emergency. He’s not a client and not on our email list.
Here’s an excerpt of his message, with a few details changed for privacy reasons and flow:
I bought property in Guatemala. It must close no later than December 19, 2025. I was told by AI that I should use a PIF to incorporate a Guatemalan S.A., and that an S.A. wouldn’t show my name in public records. I found out later that it was wrong. With the deadline coming up, can we use a pre-established PIF to speed up the process?
That was the plan. A Panama Foundation (PIF). Plus a Guatemalan corporation (S.A.). All built on an AI answer he trusted because it sounded confident.
Unfortunately, it was also a terrible suggestion. I’ll explain why in a moment.
But there’s a larger question here: Should you trust AI for international planning?
In my view, no, you shouldn’t. Here’s why.
Tools like ChatGPT or Grok have no idea what they’re talking about. AI doesn’t understand context. It doesn’t truly understand law. And it doesn’t (yet) have the ability to understand the consequences of how a complex structure works in real life.
At its core, all it does is connect words together based on probabilities, which itself is based on the data it’s been trained on. If the data is wrong or conflicting – which is often the case when it comes to cross-border issues – the answer that AI spits out will be wrong. Sounding perfectly plausible all the while.
International planning doesn’t forgive that kind of sloppiness.
Now, please don’t get me wrong. It’s a useful tool under the right circumstances. If I want to know basic facts like, say, who signed the Declaration of Independence last, I can be reasonably confident in its response:
Although, to be fair, after checking in with ChatGPT, I still looked for a reputable source to confirm, as seen here. And even then, I’m still not 100% sure. But for our purposes, good enough is good enough; it won’t get me in hot water with the IRS if I’m right on this fact or not.
But when it comes to anything that affects your money – unless you are an expert in the subject – and where actions have consequences – AI is just not qualified to give you an informed answer.
So I thought I’d take a few moments to talk about what went wrong in the example above, hopefully to serve as an example for others.
Not knowing the right questions to ask.
Most people don’t — that’s normal. But AI doesn’t fix that. It just gives you an answer based on the question you ask. It almost never challenges you to ask the right question. So if you don’t know when an answer is nonsense or you don’t know how to frame the questions properly, you won’t catch the problem until it’s too late.
An S.A. is almost always the wrong choice for US clients holding overseas assets.
A Guatemalan S.A. is a “per se” corporation for US tax purposes. That means a whole new level of compliance that’s simply not needed for a basic property hold.
And in fact, Guatemala actually has a better entity option for US owners — but S.A. is the one people find online, so AI spits it out.
A PIF is almost never the best entity to hold property outside Panama.
Although perfectly legal, outside Panama, a Private Interest Foundation (PIF) usually adds complexity. It’s a foreign structure in Guatemala from a place long considered a tax haven. That makes it harder to find professionals in Guatemala willing to work with it in the first place.
It’s done at the last minute.
International planning is not something you duct-tape together a few weeks before closing.
Apostilles, notarizations, formation timelines, bank KYC, assignments — none of that bends to a December deadline because an AI answer sent you down the wrong path.In the case of real estate, the planning should be done at least six and sometimes even up to 12 months before closing. That may sound extreme, but you might be surprised how long it takes to get things done in some of these countries.
The biggest problem? There does not seem to be any strategy.
He wanted “privacy.” Privacy from whom? Privacy in which system?
What about US reporting?
What about estate planning?
What about Guatemalan tax treatment?
AI won’t ask those questions. It gives you a (wrong) structure, not a plan.
So, ultimately, we had to turn him away. There was no time left to do it right.
And now he’s staring at three bad options:
-
Take the property in his own name — the opposite of what he wanted.
-
Walk away and lose his deposit.
-
Or get sucked into some “international expert’s” miracle fix that looks fine on a PDF but may not be completely legal.
That’s why, when we work with clients, we follow a well-worn path:
-
Figure out what’s most important to the client.
-
Figure out how to make that work on the US side first.
-
Figure out how to make it work in the foreign jurisdiction second.
-
Identify the best solution that gives our client as much of what they want as possible with a minimum of what they don’t in the most cost-efficient way possible.
And we have to start it early.
The Key Lesson
It’s not just that the AI answer is “wrong.”
It’s that the answer can put you on the wrong path, and by the time you realize it, you’re out of time.
So if there’s one lesson worth taking from his experience, it’s this: Don’t let AI plan your international life.
On the other hand, if you would like help moving assets internationally the right way – whether structuring real estate, setting up a foreign investment account, or just moving some cash offshore – feel free to get in touch and see how we can put our 40+ years of experience to work for you.
About The Author
We have 40+ years experience helping Americans move, live and invest internationally…
Need Help?
We have 40+ years experience helping Americans move, live and invest internationally…
