Divorce by ChatGPT? Why AI Can’t Replace a Divorce Attorney

We talk through the accuracy problem first: AI can pull the wrong jurisdiction, miss local court rules, quote outdated statutes, or even generate citations that don’t exist. That can wreck your credibility in front of a judge and, in some cases, lead to sanctions. We also explain why Oregon and Washington divorce practice can’t be reduced to a generic template and why “one size fits all” guidance breaks down when your case turns on specific facts and evidence.

Then we get into the part most people never consider: confidentiality. Unlike working with a lawyer, you don’t get attorney-client privilege with an AI tool. What you type and what it returns may become discoverable in litigation, requested in discovery, or even subpoenaed. We share practical ways to use AI more safely, like removing identifying details, using hypotheticals, verifying sources, and rewriting output in your own voice, while keeping the real legal strategy where it belongs: with experienced counsel.

If this helped, subscribe to Modern Family Matters, share the episode with someone facing a divorce, and leave a review so more families can find clear guidance. What’s the most surprising way AI could backfire in court?

If you would like to speak with one of our attorneys, please call (503) 227-0200 or visit our website.

Disclaimer: Nothing in this communication is intended to provide legal advice or constitute an attorney-client relationship.

Steve Altishin 0:32
Hello, I'm Steve Altishin, Director of Client Partnerships at Pacific Cascade Legal. And today we have our founding attorney, Lewis Lander, home with us to talk about why you should not use AI as your divorce attorney. Hey Lewis, how are you doing today?

Lewis Landerholm 0:47
Doing pretty good today. How are you, Steve?

Steve Altishin 0:50
Oh, this really is sort of a hot topic these days. AI is all over the place for good or bad, and it is definitely leaking into significantly not just divorce cases, but all cases. So rather than just say, oh, you you know, you shouldn't be doing this, it's really overwhelming for some people, especially if they're not getting lawyers, to have everything just like, I don't know what to do. And it's convenient to get on your attorney, get on your get on your your Facebook and search uh for some certainty, you know, confident sounding answers, and the that's depth, a chat box does that, AI does that, and it can really make you feel better. Um however, you know, uh how how really trustworthy is it if you're considering using your AI basically as as your attorney as you go through a case by yourself?

Lewis Landerholm 1:56
Uh yes. There's a lot of there's a lot of uh uh interesting challenges uh and pros and cons. And this is not a webinar about AI is bad, don't

Lewis Landerholm 2:09
use it. It's a webinar about that um AI is a good tool, but it's a tool. And it's not the um go-to answer that people you know hope that it is at times. Um, it's a AI is a good starting point, but the reality is that there are uh huge um concerns and pitfalls, especially from courts with AI usage. Um confidentiality is a problem. Um, you don't have attorney-client privilege. What you're putting into these um AI bots can be searched by potentially anybody on the internet. And so being careful about putting you know information in there that can be traced back to a specific person. Like there are a lot of challenges. You know, AI is is going so fast, so quickly, that the law is not built for fast and quick. And so it's gonna take time for the law to catch up, but you don't want to do something that could have negative consequences in your case just because you trust this bot. It's only as good as what it can find on the internet, essentially.

Steve Altishin 3:26
Yeah, and and let's say you're not hiring an attorney. One of the things that you're probably not doing is recognizing all of the different uh not just state by state, county by county, judge by judge uh interpretations and differences. Uh there they have rules, they they think things differently, and AI doesn't really capture that, especially if you don't even know what the question to ask is. Just saying, oh, can I do this in a vacuum? It's gonna come up with an answer, but it's probably not gonna be necessarily the answer to your particular situation.

Lewis Landerholm 4:09
It could be a different state answer, could be, I mean, it could be a different country answer, it could be, you know, the the the hard part, especially in family law, is that um we don't have written opinions other than appellate court and supreme court opinions in Oregon. Uh Washington does have uh some trial court opinions. Um, and so you know, that's a little more helpful. But the the challenge is that AI is only as good as what it can, you know, crawl through on the internet. And so, you know, the you know, speaking of the Oregon side, there just isn't information out there that is specific to every single situation. Like you said, every judge has their own way of doing things. Courts have they have their trial court rules, they have their state rules, they have unwritten rules, they have all of these different things that you know you have to learn that can only, you know, a lot of them can only be learned through experience and time. So it's not gonna be a fail-safe and a foolproof exercise to go to AI to ask, say, you know, how do I file this in, you know, say Marion County, Oregon. You're not going to get, you're gonna get uh a basic like go to this website potentially, but it's not going to tell you how to do something. And then there's all the nuances that just get really challenging with how judges are gonna rule, what facts matter, how to deal with evidence, all of these different things are just, you know, they're complicated, and and AI isn't necessarily going to be able to answer those questions.

Steve Altishin 5:51
And, you know, like you said, it's not necessarily we're saying AI is bad, but there are still cases out there where you just get fake information. And that can really be hard in your case because if you hand a judge this is what you know, this is what I should be getting, when it's using fake information, they're not going to be happy at all. You can get in real trouble.

Lewis Landerholm 6:20
Yeah, they there are situations where people have cited cases that gave their position authority, both attorneys and unrepresented parties. Judges have responded by actually, yeah, issuing sanctions and fines against those parties. It doesn't mean that you can't start by asking the question to AI, but if you don't check the citation and the quote to make sure that it's actually a real case, number one. Number two, that it's a case that stands for what you're saying that it stands for, judges get prickly about those type of things. And so you have to be careful that we're looking at a statute that still exists. A lot of times a statute or a or a law gets amended or or overridden, but you may it may not be easily recognizable when you look at it. And so AI is going to find maybe the first version, doesn't you know that they come across, but maybe not the amended version. And so it's a tricky thing to deal with, but just means that you have to take some time. Also, have having representation who has the experience of how to do these things and and how to argue, also what to look for is very important. AI isn't going to replace litigation anytime soon just because the of how constraints are, since it's it's bound to what it can find within written word, you know, just not everything's going to be able to be extrapolated from that.

Steve Altishin 7:51
And and I know at least a couple of judges I've talked to who have talked about the growing amount of erroneous sites, erroneous arguments, erroneous sort of everything. And it's starting to pile up the judicial system. I mean, it's just they're having to sift through it. And and if you're gonna get a somebody who cites something from Nevada and not Oregon, that judge is probably not just gonna say, Oh, that's no good, but they're gonna put a kind of an eye on you and go, Can I believe anything you're saying?

Lewis Landerholm 8:36
Yeah, yeah, and that's the challenge right now, is that figuring out where everything is coming from is is definitely it's not easy. It's not a an easy exercise to be able to punch in all the stuff into the computer, have it spit out everything for you, and then just take that verbatim to court. Doesn't it doesn't work that way.

Steve Altishin 8:58
Because they sort of sort of use the average, they amalgamate and come up with a with a an average. Now, the other the other thing about AI that I think I might I'd like you to talk about is not only it being inaccurate at times, and maybe at the worst of all possible times, but it's not an attorney, and it doesn't have the advantages of having an attorney because it just it's it's not.

Lewis Landerholm 9:26
Well, and that brings up confidentiality and attorney-client privilege. So there are instances where opposing counsel and opposing parties are asking for uh searches within AI bots, and so you're required to turn over what you were searching and those answers. So that's a big concern and uh big risk in using some of these AI products that are out there. The other thing that you know you have to think about is because AI is going into this public domain and you know it's a third-party search tool, those can be discoverable. So, you know, whether you turn it over or not, the other side could subpoena the company to turn that over. And then also other people who are searching within AI can find your searches and answers as well. So there's a lot of pitfalls that go along with using AI as your attorney. Whereas when you're working with an attorney, we can't be forced to testify about what we talk about. And you as the litigate can't be forced to testify about what was said as well. And so there's a you know, that's a that's a big deal when you get into court so that you can trust that that communication flow is not gonna come out end up in front of the judge.

Steve Altishin 10:56
Yeah, and you, I know when you talk to your our clients, you want the good, the bad, and the ugly. You want to know everything. And much of that stuff, if it goes any farther, could you know be could be bad for your case. And AI doesn't have that protection. If you put in, oh, I just I may have pushed my my spouse around a few times and you plug that in, what am I gonna get? And it'll say something, and you know, oh you've now put that out in public for all to find out.

Lewis Landerholm 11:31
Yeah. One, you shouldn't be pushing your spouse, but two, yeah. Yeah, yeah, that's not that is the main reason. Right. Putting uh putting all of that out into the public domain is not a good look for anybody, and it can definitely affect your case if you're asking for custody or asking for some of these issues that come up in court, or even if there's a restraining order or a criminal case comes down, like you you want to be very careful about what you're putting into any sort of search engine.

Steve Altishin 12:07
Yeah. So we talked about this thing that that's sort of starting is the attorneys are getting wise to this. And if it feels like it's the sort of computer-generated stuff, they can put a request for that in their discovery, request for discovery, and you can't say, no, I mean, this is private, this is my private stuff.

Lewis Landerholm 12:31
Yeah, and this is where the law is trying to catch up to the technology. There are cases out there that judges have ordered and said that it's completely discoverable and you have to turn it over. And so this is going to be sort of a process, and things are going to take some time for it to shake out on how all courts deal with um AI usage and what's turned over and what's what's required to be turned over in cases, but it's there's definitely risk there, and you know, you can't assume that just because you're using an AI search that you're not gonna have to, you know, reveal to the other side what you were typing into it or what it returned in an answer.

Steve Altishin 13:18
I imagine because I know you get a lot of clients and the other side a lot of clients who want to go on just social media in general and rate their ex over the sp over the calls and all of that. And AI can be the same kind of thing where you know, oh, if my my spouse has done da-da-da-da-da-da, what what would happen? I mean, and it feels like it just it's too easy to get caught doing things you shouldn't do.

Lewis Landerholm 13:49
Yeah, absolutely. Yeah, I think that's a good analogy, is like anything that you wouldn't want to post in social media, you also shouldn't post in an AI chat. I I think that's a a good good advice and and a good way to think about it, that it's in the same realm as social media, that it can be just as public and people have a sense of privacy there in the chat bots, but I think that's a misguided trust of it, and it can definitely be a problem

Lewis Landerholm 14:26
in a case. And even if we back up to even more, you know, sort of mundane use it and lesser issues than it being like discoverable, just the fact that when you're using AI, it's you know, it doesn't have your voice in it when you're writing. It doesn't have just the way that you would explain something. It's you can tell when something is is just thrown into a chat and then people are using those responses. So say again, using AI is not a bad thing. Make sure there is no identifying information that can be traced back to you specifically. Uh, a lot of people will use hypotheticals within AI so that way it's not have the discovery type issue nearly as much because it's not you putting in your own facts necessarily. They can still discover that information, but it's not like you're revealing a lot of you know information that would be detrimental to your case. But then from there, when you get the answer, don't just turn that in or don't just send that over verbatim. Go through and actually like rewrite it, put it in your own voice, in your own words, make sure it's edited, make sure it actually sounds like something you would say and not something that computer or a uh a robot would say. And then AI can be a positive tool to help in the right direction. It's never going to replace the advice component, though. The thing that attorneys and people in the legal system do, what our job is, is essentially to have years worth of factual fact patterns that we have seen, and we have seen how the courts deal with those fact patterns and how they rule in response to a certain set of facts, and then come up with the answer. And so, what we do as attorneys is we've we've seen thousands of those particular answers that have been given by different judges, different counties in relation to the issues that go on with cases. And then we can help to guide and advise our clients on in your case, this is a realistic number, this is reasonable, this is how we would approach a case where AI doesn't have that same level of ability to have that experience level. It's maybe way faster at crawling a document for sure, but it doesn't have the ability to take from multiple sources, whether it's written, whether it's verbal, whether it's in person just by words, whether it's from another attorney who said that this is what happened in this county. Like all of those things are what we rely on, you know, over the over the years of practice to be able to guide and advise our clients. So that's the piece people miss out on if they're relying to have on AI.

Steve Altishin 17:33
The legal research part, it feels like, is where it goes kind of south because they don't have you to then form strategies. And they can give all this information, but then eventually someone has to take that information and form a strategy to either try to get it settled or go to court, and then they're just back to themselves again. It's just because the strategies that AI sends out are again just sort of populous, they're the one size fits all strategies.

Lewis Landerholm 18:09
Yeah, the actual strategy part is going to be missing. It'll get the facts right, but it's not going to be able to then make a well-informed, executed strategy on how to uh move forward.

Steve Altishin 18:22
So what would you say to a client who comes in or someone who's maybe going to be a client and says, Well, I've ran this stuff through AI, through the other kind of chat boxes, and I think this is what we should do. Is that starting to happen to you? It's like, well, hold it. AI said I should do this. Why are you saying I should do that?

Lewis Landerholm 18:46
Yes, we're definitely seeing more and more of that. We try to have the beginning or the conversation from the very beginning about that AI, you know, use it for the tool that it is, but don't rely on it for the answers so that we can talk about it. We want people to be able to have an understanding of their case where they feel like they can go and search potential answers, obviously, with keeping the confidentiality piece as well. But to be very clear that one, our job is not to just do what AI tells us. Our job is to tell you what we think and what we advise you to do. So it's not a great place to start in a case if you're telling legal staff what they should or shouldn't do based on AI. It's not going to start the relationship off in a positive way because AI has limitations and it's not going to go over well with anybody who's representing you if you're saying, well, AI told me. It's one thing to say, well, these are some of the things I found. What about this? And asking questions, but just to dictate to do it as AI has said, that doesn't work with the courts, doesn't work with the professionals who have been representing people for 20 years, 30 years, 10 years, that that's just irreplaceable experience that you can't replicate.

Steve Altishin 20:11
Absolutely. Yeah, this is your skill. This is this is where you come in. And that even it seems like because AI is now, I've seen doing tax issues and psychological, professional issues, like trying to, you know, you have this, your spouse has that. It's not good for those either.

Lewis Landerholm 20:31
Yeah, there's definitely yeah, people need to understand where to use AI and where not to use AI. And that's that's just a part of the natural progression of technology, and it's gonna take some time for everybody to be able to understand that.

Steve Altishin 20:46
Yeah. Well, we blew through our time, Lewis. Thank you for sitting down again to talk to us about why you should not use AI as your divorce attorney. Not saying don't use them, but don't use them as your divorce attorney because it's just too complex for AI right now. And I think that's great information you talked about. So thank you again.

Lewis Landerholm 21:10
Yeah, you're welcome, Steve.

Steve Altishin 21:12
Oh my gosh. And thank you everyone else for joining us today. If anyone has further questions on today's topic, you can post it here. We can get you connected with one of our attorneys. And until next time, stay safe, stay happy, and be well.

Intro/Outro 21:28
This has been Modern Family Matters, a legal podcast focusing on providing real answers and direction for individuals and families. Our podcast is sponsored by Lander Home Family Law and Pacific Cascade Family Law, serving families in Oregon and Washington. If you are in need of legal counsel or have additional questions about a family law matter reported to you, please visit our website at landerhomelaw.com or Pacific CascadeFamily Law.com. You can also call our headquarters at 503-227-0200 to schedule a case evaluation with one of our states' attorneys. Modern Family Matters, advocating for your better tomorrow and offering legal solutions important to the Modern Family.