Artificial Intelligence and the Unauthorized Practice of Law

This episode examines whether artificial intelligence tools, including ChatGPT, may cross into the unauthorized practice of law (UPL).

The discussion highlights a recent federal lawsuit, Nippon Life Insurance Company of America v. OpenAI, where an AI platform allegedly influenced a claimant to revive previously settled litigation.

Visit our website for information about our legal services related to emerging technologies.

Transcript

Copy Transcript Copied

REMINGER REPORT PODCAST ON EMERGING TECHNOLOGIES

Kenton Steele, Esq.

ZBP       Zachary B. Pyers, Esq.

KS          Kenton Steele, Esq.

 | Speaker | Narrative | ZBP | Welcome to the most recent episode of the Reminger Report Podcast on Emerging Technologies.  I’m Zach Pyers and I’m joined today by my partner, Kenton Steele, also the co-host of the Reminger Report Podcast of Emerging Technologies.  It’s a mouthful.  And we’re going to be talking about the unauthorized practice of law and artificial intelligence.  And I know, cause I just read a recent article in the American Bar Association that talked about how artificial intelligence has kind of been a spring board to pro se in litigation and has increased both the number of filings, but also the depth and the length of those filings, how long they take to resolve.  And so today we’re kind of looking at a unique approach which is, does artificial intelligence actually cross over and cross into the area of the unauthorized practice of law.  So, Kenton, with all that said, thanks for joining us here today.
| KS | Yes, I’m very happy to be here.  It certainly is an interesting topic and have seen myself the impacts of artificial intelligence on pro se litigation and can certainly understand how the increase in the cost of defending cases filed by pro se litigants using AI.
| ZBP | And so today we’re actually talking about kind of a unique situation where there actually was a lawsuit filed challenging an artificial intelligence company that suggested that their platform constituted the unauthorized practice of law.  So, why don’t you just give us a little bit of background about that.
| KS | Yes, it’s a very interesting case that was filed in federal court in Chicago.  The suit was filed by Nippon Life Insurance Company of America.  It was filed against OpenAI, which is the operator of ChatGPT.  The thrust of the lawsuit is the insurance company accusing OpenAI of the unauthorized practice of law.  Now the case arises from a dispute that was ongoing between the employee of the life insurance company over long term disability payments.  There was a prior lawsuit about that case that was settled between the insurance company and the employee, but the employee then went to OpenAI to ask if the information the employee was getting from her lawyer was accurate or if she should do something different, challenge the settlement, and as a result of the information the employee received from OpenAI, she decided to file a pro se complaint attempting to revive this previously settled case, this dispute about the disability payments.  In the course of that litigation, between the employee and the insurance company, the employee filed tons of motions, lots of different re-filings were submitted, and ultimately that case was dismissed.  But now the life insurance company is suing OpenAI, accusing it of having engaged in the unauthorized practice of law with respect to the advice that the employee received from OpenAI.
| ZBP | Now one of the things that we kind of talk about, at least in the legal profession, right, is that there is a lot of legal research fools out there.  I mean, there’s the commercial ones, there’s free ones.  There’s lots of legal research tools.  And then there’s also kind of this other side, right, which I would categorize that are outside of legal research tools, we also talk about legal advice, which crosses a line right into potentially practicing law.  So, how do we distinguish or how do you draw that distinction between a general legal research tool and a tool that might actually cross the line into practicing law.
| KS | Yes.  The real distinction has to do with whether or not general information is being provided or you are just retrieving precedent or a statement of how things were versus receiving advice on a particular situation.  That’s really way on-line crossing habits.  So, you would do all the research you like to and read about other cases that were decided, but as soon as you cross that line into interpreting how do the cases that I assign apply to my particular situation, which is what the AI chatbot is accused of having done here and likely did.  That is typically reserved for the practice of the purview of lawyers.  Right, that is practicing law once you apply the information that is available to a specific situation.  And in this case, it even goes further than that where you have the AI chatbot actually drafting legal documents to be submitted and filed.  So, we have two steps beyond just the basic research tool where information got into actually providing legal advice and preparing legal documents.
| ZBP | Now one of the things that I think is kind of, I don’t know, let’s say unique but is interesting or stands out – it’s kind of the background of the litigation that you talked about as – the case was actually settled and dismissed, and after the use of the AI, the case was reopened.  And so how does that allegation almost, the AI platform, encourage, which is kind of unusual, right, I mean you and I both are litigators that settled a lot of cases, dismissed them, it’s not common that we see a case reopened after that stage.  So, how does that fact kind of play into this whole argument surrounding the unauthorized practice of law?
| KS | Yes.  So I think you have a situation where because of the advice that this former employee was receiving from the AI chatbot that the employee was talking with, this lawsuit was filed.  I think you have a clear case of OpenAI in some ways being the but for cause of the case continuing.  You know, the employee had a prior attorney who had discussed with the client what the outcome should be, reached a settlement and the case was over, but then the employee decided that she was unhappy with that result or was unhappy with the information she received from her lawyer.  Now, that’s where the AI chatbot comes in.  And one of the things that’s interesting here is it’s in the nature of the chatbot to affirm what the person is asking.  So if the person, this employee, comes to ChatGPT and says, I think my lawyer’s giving me bad advice, what should I do about it, the chatbot is unlikely to tell you, just listen to what your lawyer is saying, don’t ask me any questions about why this woman could promote ways for you to continue to engage with it, engage with the chatbot, create further feedback and incentive to use this chatbot, which furthers the cost that’s ultimately borne by the insurance company with the plaintiff in this case.  So, that’s likely where you create an argument that the chatbot was the cause of the additional legal cost paid by the insurance company in addressing this previously dismissed lawsuit because no attorney presumably would have told this employee to try to revive this litigation and pursue it further.
| ZBP | Now, obviously a lot of these entities, I mean in this case there actually wasn’t an insured involved, but when we think of traditional litigation and we think about often times there is an insurer involved on the defense side, you know, kind of ensuring the defendant -- what type of challenge does this firm present for those insurers who were encountering  litigation on a routine basis, when we’ve got these AI assistant filings.  I mean what risk does that present to them?
| KS | Yes.  It is something I have encountered in my own practice and seeing what other people have to deal with.  I knew that early in my career, pre-AI, pro se litigants – and this is not all pro se litigants, let me process for saying AI.  There are certainly a number of pro se litigants in cases who are able to represent their position in the case capably, on their own behalf, and put forward things.  But there is also a certain percentage of pro se litigants who don’t know what they are doing, who are not able to go through the steps of the legal process.  In early of my career, we would see pro se litigants file cases that really had no merit whatsoever and typically it was very easy for a court to bounce those cases out because the plaintiff had failed to follow some procedural hurdle.  They didn’t submit their filings in the right way, things were not formatted or templated the right way and the court came, before having to even get into the merits of what the plaintiff was saying, particularly if there was some outlandish plan, the court could just sidestep that and say, oh, you used the wrong font or this is not, you know, the margins are wrong and so we won’t consider this.  You can try to refile if you want, and if you don’t refile in 14 days, we’re just going to dismiss the case.  And that is how lots of pro se litigation ended, was with the plaintiff running into a procedural hurdle that they were not able to overcome.  With the advent of AI chatbot that could draft documents, you can upload the local rules of the courts to the chatbot and ensure that all the filings comply with all of the formatting requirements.  In addition to that, the submissions that are drafted by AI – I have a case going on right now in federal court against a pro se litigant where it’s patently obvious that the guy is using a chatbot to draft his legal documents.  The problem is they look legitimate.  Right?  They match the form of a legal argument, even if there is no substance there, and that creates an additional burden for having to respond.  The joke is that we used to see pro se complaints that were like written on a sheet of notebook paper with a crayon and it was very easy to just go denied and the case would get kicked out.  But when it was getting looks like it’s right, you then have to deal with the substance.  And additionally, I have seen another case where the pro se claimant where the number of filings was really high.  The person would go to the chatbot and ask questions about how to do something or how something should happen, and it will tell the person, we’ll file a motion about it.  Many times it’s things that are handled at like a prehearing conference. Or something that gets addressed in communications between the attorneys rather than a motion being filed about it.  But the pro se litigant doesn’t know that.  And at every time something gets filed in the case, both the court and opposing counsel have to review that filing, determine if a response is necessary and all of that results in additional cost to whoever is footing the bill for the defense in the case which is often going to be an insurance company.  So, I mean that’s where the big problem is with litigants using AI to assist with their filings is they may be told to file more things.  Those things are going to look more polished and require more of a substantive response than pro se litigants filing in the past.  And we have seen some changes from OpenAI and some bills being considered that would just prohibit the drafting of legal documents or giving of legal advice.  OpenAI purports to have a policy now that wasn’t in place at the time of this case we’re discussing that should prevent the chatbot from giving legal advice.  I have some doubts as to how effective that restriction is.  There’s a lot of examples of people finding ways to get around the chatbot, say they’re not going to do.  So I think that’s certainly something that can be overcome and then once we reach the point of laws being passed about this, which New York is loading the ball apparently to prevent the giving of legal advice or medical advice.  We would get a really interesting outcome at that point.
| ZBP | So, we took interest in this litigation.  Open AI, who is the defendant and has essentially said the claim lacks merit, or I might be reading more into it, but that’s the general. When you look at this from a defense perspective, you are defending OpenAI, what arguments are you most likely going to be trying to latch on to say, forget the bills and the laws that are being passed and the policies – even at this time when this was going on, this didn’t necessarily constitute the unauthorized practice of law.
| KS | Yes.  I mean, at bottom, nothing was filed by OpenAI.  Right?  OpenAI wasn’t representing this Ms. Livingham.  She was generating things and making the decision to file it on her own.  So, while OpenAI may have provided the resources and put things together to be filed, it likely doesn’t meet the technical definition of unauthorized practice of law for that reason.  There wasn’t – AI didn’t make an appearance in court.  It didn’t file anything.  The litigant made all those decisions on her own.  That’s where OpenAI likely hasn’t been a strong defense to this case because it is a very novel approach the insurance company is taking here, likely sending a message more so than winning on the merits of this case, in my estimation.
| ZBP | Now we talked about some of the disclaimers.  We talked about so many policy changes.  We talked about some of the laws that people – states, various states, are looking to implement.  As AI companies move forward, do you think there is anything else that they should be doing?  I mean anything else that you know to prevent this type of risk, even if it’s, like, I mean even if the risk is small because it ultimately may not be successful?
| KS | Yes.  It’s tough to say.  I mean I think one change that has already started to happen that we have seen with presumably Open AI is better tuning of the AI software of the version of the chatbot as we discussed and the burden of ChatGPT that was at issue here is likely to ChatGPT 3 or 4, which was somewhat notorious for being a yes man where they would tell people what they wanted to hear and tell them they were great and special and it would get people to continue to use the chatbot.  A better tuned interface or platform that would tell someone, no actually you don’t have a lawsuit, don’t waste your time would be better.  And that’s sort of a global change that could happen and we are seeing happen with the large chatbots that are out there.  Beyond that, I think, you know, placing those internal restrictions in place as to prevent people from asking for legal advice or drafting of documents is a good idea, even if there is likely to be some flaws.  And I think also creating systems where like when a person is asking for legal advice, they should be directed to their local bar association instead of receiving any further answer, right.  Just like if a person seeking medical advice, give them the number for 911 or the information to book a doctor’s appointment depending on the severity of the situation rather than answering the question, I think it would probably be the best solution.
| ZBP | Now, kind of as I wrap this up, you know, we’re legal practitioners.  Obviously, this is how we make a living is practicing law, representing clients.  What practical steps do you think that we should be taking, right, as we look to the future of AI when we are both defending claims that are being prosecuted with the use of AI, whether they are pro se litigants, whether they are people using AI to assert claims against us or other practitioners?  How should we be thinking about this risk and then how should we be approaching it?
| KS | A couple of things.  I mean, on the one side, and this is the advice I have to give clients frequently is they ask me what do I do to keep myself from getting sued again.  The short answer is nothing.  Right.  There is nothing that you can do that will prevent a person from filing a lawsuit.  All you can do is position yourself to win a lawsuit in the event that it is filed.  And so, I think a lot of the response there is going to be your traditional advice of legal practice, good business practices are your best defense against a lawsuit being filed.  I think one thing that clients should use more frequently, especially if there is any type of contract related, is something you intend to is fee shifting provision, which can be an especially potent tool when you’re dealing with a meritless claim, to be able to go back to the plaintiff and say, if you lose, you’re paying my legal bill for this case and you should really carefully consider whether or not you think you’re likely to prevail.  And I can help defray the cost, whether or not they are recoverable or not, as kind of a separate question.  You still generate the leverage of being able to make that argument. And so it’s something that clients who are in a position to be sued by a person claiming using any AI chatbot should be considering adding those terms to their agreements to put himself in a better position in the event a lawsuit is filed.
| ZBP | You know, so I haven’t heard the fee shifting and it’s interesting, right.  So for those of you who may not be familiar with the American rule in America, generally don’t get attorney’s fees even if you win the lawsuit.  Generally, each party bears their own cost of litigation, including their attorney’s fees.  So, even if you are successful, you still end up losing whatever amount that you or your insurance carrier had to pay the lawyer to defend you.  So, what you’re suggesting actually is contrary to the American rule but use in other jurisdictions outside the United States which would essentially shift the cost of the litigation to the other party, which would be an interesting approach when you think about it because it may limit some of the frivolous lawsuits.
| KS | Yes.  Create and lose your pay system I think could be beneficial to help deal with a situation like this.  Now, we have always had the American rule to create easier and open access to courts.  That type of agreement would run counter to that, but we also have freedom of contract here, and so if that’s something that you can give clients to contract and accept, it could be a worthwhile thing to do it.
| ZBP | Yes.  That’s interesting.  Kenton, I appreciate you taking the time today to talk to us about this case and it will be interesting to see how it ends up resolving.
| KS | Yes, absolutely.  Happy to come back once we have a decision here.
| ZBP | Sounds great.  Thank you, Kenton.
| KS | Thanks.