Your AI Chat History Is Discoverable in Court

Collin McKee

Welcome back to the Endeavors AI Podcast. I'm your host, Collin McKee, founder of Endeavors AI, where we talk about the intersection of artificial intelligence, law, business, and the new economy being built all around it. My guest today spends her days on the exact problem most business owners don't yet know they have. Jennie Woltz is a labor and employment attorney with about two decades of experience in the field. She co-founded her own firm, Woltz & Folkinshteyn, and she advises employers across New York, New Jersey, and Connecticut on the unglamorous stuff that decides lawsuits — handbooks, leave policies, hiring practices, performance reviews, and the like.

Here's why I wanted to have her on. Your HR team is likely using AI today. And the risk isn't that it writes something wrong — it's that it writes something that looks completely right and is quietly missing the important part that might keep you protected. Jennie, welcome to the show.

Jennie Woltz

Thanks so much, Collin. I appreciate being on.

Collin McKee

Before we get into the AI piece, I want to start with you, because you built something I think is genuinely unusual in law. And you built it before AI was part of the conversation at all. Give me the two-minute version if you would. Twenty years in employment law, and then you go and start your own firm. What made you do that and go out on your own?

Jennie Woltz

Great question, Collin. I saw, from working both in and out of large firms and in the nonprofit area, that there was a huge need for someone who could come into businesses and help reduce employee drama, streamline HR, and give people the quick, speedy advice they needed at the price they could afford — to make sure their workforces stayed drama-free and risk-free from lawsuits.

I just felt that was a really important goal, and it was something I really wanted to help people with, and help the world with. So together with my partner, we formed our own law firm. It's very much built around the idea of providing proactive advice, counsel, and solutions that help prevent the lawsuits before they get started. And then of course we do all of the regular things that a law firm does — defend and prosecute claims as we need to, to keep employers risk-free. We want to keep the workplace a dignified place to work for everybody, whether you're an employer or an employee.

Collin McKee

Let's go into a little bit of the detail of what makes this unique. During COVID, you built an employment law co-op. Walk me through that. What was happening during that time? Who was calling you? What was the catalyst that really made you think the hourly model wasn't going to work for them?

Jennie Woltz

During COVID, as everyone here can remember, the world we lived in was rapidly changing. I mean, the idea that we're doing this interview over Zoom is just one such artifact of that. But the workforce in particular had so many changes happening — between masking laws and vaccination laws and new sick leave laws — that things, especially in the tri-state area where I practice, were changing constantly, and employers had so many questions.

A couple of my clients came together and asked me and my partner, who's also my husband Ben, if we would consider doing a regular weekly or biweekly seminar for them, just on the changes, so that we could help their HR departments stay up to date and present on those ever-changing laws and deal with those issues as they were coming up in real time.

And what started off as a very collaborative group of a couple of employers pooling their resources and getting some real, boots-on-the-ground, practical legal advice ended up blossoming into what we now call a co-op of legal services. We now have eleven or twelve different employers who are part of this employment law co-op. What happens during this co-op is that employers can come to us with any sort of day-to-day question. There is a joint representation agreement. We try to keep the questions anonymized, but we will share the questions asked and the solutions presented, so that the twelve employers can get essentially twelve times the legal services for the price of one.

Collin McKee

Dig a little further into that, because this is a new topic for this show, and probably for most of our listening audience. How does the joint representation agreement actually work? I feel like a lot of people think that it wouldn't, but you've made it work.

Jennie Woltz

Everyone knows who's part of the co-op. So they know that if they hear a question I'm presenting — "hey, one of you asked this question" — it's one of the twelve, right? It's not an anonymous person. It's a real-life situation that some other HR person within this group had. And the joint representation says, listen, this is privileged for all of you.

Again, when we present the topics, we won't say who it is, but inevitably, in a group setting, there'll be one HR person who'll say, "that's my question." And so people keep it confidential as if it were their own.

Collin McKee

Yeah, it makes perfect sense. It's just outside of the norm, so a lot of people wouldn't go there by default. Another thing that jumps out at me is that it's been growing year over year for you. So it's obviously working, and it's something that's a real demand and real value for your clients. Was there a moment where you realized clients weren't buying your hours anymore, they were buying solutions — and was that the catalyst to change?

Jennie Woltz

That's a good question. I want to be really clear: we have plenty of clients who still pay us hourly. Although I will say that many clients have also wanted to start saying they're interested not just in — they don't care how long something takes. They want a solution. They want something that's more value-based. And there are certain kinds of projects that lend themselves to that very well.

If we're redoing a handbook, or we're doing some kind of discrete project where there's a clear outcome, a clear deliverable, it makes sense from all perspectives to get some cost certainty. The last thing clients want is to be surprised by a bill they weren't expecting. If they can get the value and we can all agree on a price, then that's good.

But the other thing about working on a fractional basis, or on a subscription basis over a long period of time, is that it's one of these things where, once employers have it, they realize how much they need it. Reducing the friction is one of the biggest benefits — if you've already paid for something, if you've got a subscription, it reduces that friction. So HR has no reason not to call you for a question. And so you as the employer can really rest assured that all the decisions being made day to day are helping your business be quicker, more efficient, more drama-free, more consistent — and you're reducing risk while you do it. So it's a benefit to everybody.

When you're paying hourly, there is a tension between bettering your business and feeling like it's going to affect your bottom line. But you don't have that same concern when you've already budgeted for these kinds of preventative legal services.

Collin McKee

That's a great segue. You said a couple of words that stick to the next topic — decision, benefit, and liability. In the topic of AI, a lot of people think they need to make a decision on how to use it, but likely decisions have already been made and people are using it, and they're looking to implement it to create efficiencies and to avoid liabilities.

Most likely, everyone's HR person or department is using a chatbot in some way, right? It's just not necessarily in a policy, it's not structured just yet. So give me some real-time examples that you're seeing where AI is actually showing up in HR functions and people are surprised by it — where they don't realize there are legal consequences. That's the meat of our discussion.

Jennie Woltz

Yeah, one hundred percent what you said. And you said something at the beginning of the episode too, which I thought was really important — that AI ends up creating this sense of certainty with whatever you're doing, and it simply isn't there.

So where are we seeing this in HR teams? A couple of different places. The first place is in adopting systems that use AI to screen applicants. There's a very direct potential legal liability in all of these screening tools, which several states have actually started to address and legislate. So if you're in a jurisdiction that has an AI screening tool law, you're obviously wise to look up what that law is and make sure you're using the screening tool correctly.

What are some of these laws — in, for instance, New York, Connecticut, a couple of other places? They basically say that an AI tool — and they're different in the different jurisdictions — either needs to be disclosed to the candidate, or it needs to be vetted for discrimination risk, or it needs to not be the hundred percent solution for screening. There needs to be a human behind it. And sometimes it's all three.

So I think screening tools are the biggest immediate risk that HR right now is using AI for, and it could lead to some pretty bad problems. The real concrete risk there is if you're using an AI screening tool and it's creating a disparate impact — your screening tool for whatever reason is only selecting people of a certain race, gender, color, creed. There's no defense. There's no legal defense to say, "well, it was just the screening tool, it wasn't me." You are responsible for all of your AI tools.

The second place we're seeing AI come up quite a lot is employers trying to use AI for legal or quasi-legal documents, such as handbooks or contracts. And while, as you pointed out, AI might come up with something that looks good — just because something looks good doesn't mean, A, it's legally compliant with your jurisdiction or your employer makeup, or that it doesn't have gaps.

That's oftentimes because you can put into ChatGPT or Claude, "hey, make sure that this is a handbook that complies with X, Y, and Z," or a contract that complies in this jurisdiction. And so it might be right, but it's not complete.

Inevitably — even when I'm using AI tools in my own practice to get the ball rolling on drafting and things like that — it's really about looking at the chess move three moves down. What's going to happen if this other thing happens? What happens if this risk is there? Is the way this is drafted actually protecting me, or is this just a generic clause that was put in there? ChatGPT, Claude, they all make assumptions about what you as the drafter want. And sometimes they don't even know they're making those assumptions. So there's a lot of incompleteness.

And to be fair, even before AI and ChatGPT — you had mentioned, and it's true, that one of the things my firm does is review handbooks for compliance. We used to see all the time that employers would take templates from their friend's company, or something they found on the internet, or whatever it was. Here's a really great example: a tiny employer who would have an FMLA policy. But FMLA doesn't kick in until you have fifty employees. So sometimes things would be in there that didn't need to be in there. Sometimes things would be in there that weren't sufficient. Employers would say, "well, this is the way we do our PTO policy," and we'd have to say, "but there's a sick leave law that guarantees X, Y, and Z."

Anyway, the point is that even before AI, if your HR team doesn't have the judgment of what needs to be in that handbook or not in that handbook, you won't get the results you want from AI. And this is especially important when it's not just handbooks or employee-facing policies, but contracts — which of course are meant to be written in the event they need to be enforced in a court.

The third area where I think we're really seeing a lot of AI use is in general employee communications. Good HR people know that every employee communication could at some point be exhibit A in a future lawsuit — how HR failed to let you know about a leave right, failed to escalate something appropriately when you complained, failed to give you proper information, or any number of different things. So good HR people recognize that every email that gets sent could truly be an exhibit later.

But when HR people are just trying to manage their inbox of a hundred and fifty emails a day and respond to people, and they're using AI tools to quickly respond to the issue, they will miss larger-picture issues. And they may not be saying the right things. So too-quick use of AI to respond, without using real knowledge and insight of what the overarching picture is — what the overarching concern that employee has is — can end up creating a lot of missed opportunities. For instance, to make sure the employee is being offered the leaves they're entitled to, or to make sure you're starting your interactive process if an employee has mentioned a need for time off or some kind of disability.

It's really reading between the lines. That's the art of HR. And AI doesn't have that same judgment that a seasoned HR person — and certainly an employment attorney who can see the lawsuit waiting to happen — has.

Collin McKee

In our first call you likened it to an 80/20 rule — that AI can get about eighty percent of it spot on, but there's that twenty percent that it might fill in, and it's going to look right, but it's that twenty percent that can get you into a lot of trouble. Why does that twenty percent need so much focus from a professional? That twenty percent that gets into the intricacies, like you mentioned with FMLA, makes it quite obvious that you didn't really do your research, or you didn't have guidance to do this the correct way. And it can really get people in hot water.

Jennie Woltz

Right. The twenty percent you mentioned — that's why you're hiring a professional. To be able to see those things that maybe eighty percent of people don't necessarily see. You want that expertise in your business because you don't just want to be middle of the pack, right? You want to be seeing around the corners. And it's that twenty percent that AI is not seeing around the corners.

What AI does is it takes what's out there already and averages it. It's not looking at what the case law is. It hasn't been to litigation and had to defend its decisions against a jury and against evidentiary standards. Using someone who has been there, and has seen it, and can anticipate those risks and then better set you up for success — that's really where a lot of the art comes in.

Collin McKee

It definitely does. Another thing to coattail on that: AI can do it if it's given all the proper guidance from a topic expert. But if you don't know what to ask, then the AI just grabs at whatever to fill that space. Someone like Jennie could probably use that tool and ask it all the right questions and make sure it's covered. But for the layman who doesn't know all the case law and all the ins and outs and intricacies of what the new laws being written are, that can leave you rather liable, and you could find yourself in some trouble.

Business owners listening to this — we don't want to scare you, we don't mean to scare you, we want to bring you information. But you're right to be nervous and to want to address it. So let's talk to that. If I'm a business owner and I'm trying to stop my team from creating these liabilities for me, I'm not going to be able to tell them "hey, don't use AI." That ship has sailed, everyone's using it. What can we do to be responsible with our employees, to actually have responsible AI use? What does that look like for a firm, or what do you recommend to your clients?

Jennie Woltz

My advice to firms is a couple of different things. First, accept that your employees are using AI, and have an AI use policy.

One of the things we actually didn't talk so much about on this call — it's not specific to employment law or litigation at all — is employees using AI behind the scenes. If it's not firm-sanctioned, they're likely putting in your company's trade secret information, your donor details, into some free ChatGPT or some unpaid Claude subscription, and who knows where that's going.

Have an AI policy. Accept that people are using AI. Ask them how they're using AI. Put limits on it. And make sure they're not putting your business's confidential trade secrets, confidential client relationships, things like that, into some box that you have no control over. So have a policy, and that policy should dictate how people should use AI. I think that's the number one thing.

The second thing is: get clear on how people are using AI, so that you as the employer can identify what sources of risk are happening. If your HR team is using AI to communicate with employees, in what way are they using it? Are they using AI to substitute for their own judgment? Are they relying on AI as a false sense of security, when you should have true expertise before using that for things?

Are you adopting new technology? Going back to one of the risks I mentioned earlier with some of these AI selection tools and the AI selection laws — if you are using any kind of AI selection tool in your business, check. See if there is a law that directly affects that tool. And even if there's not, it's best practices: audit that tool and make sure it's not creating a disparate impact risk for your company.

Disparate impact, again, is the idea that you can have a tool that has some internal bias in it. And — surprise, surprise — if you are using your existing workforce's data to influence your AI decision-making tool, there is some bias in there. I don't know what it is. I can't tell you sitting here if it's good bias, bad bias, or not. But you should have it audited and make sure that you are not perpetuating biases that already exist in your workforce.

Collin McKee

That's great advice. And good for you — you got to this topic a beat before I did. That was actually going to be my next question. But yeah, use the tool for what it's worth. No one wants to dig into the terms and conditions of everything we click yes on, but when it comes to sensitive information, for tools that you're using for your workforce, it's worth it. It's worth scanning it, and giving the tool your municipality so it knows what your local laws and rules are per state. It's an easy and perfect use of these tools — that's what they're designed to do, that tedious work of large document review, just to cover yourself.

Another thing a lot of people haven't put into place, and the discussions are starting to happen, is an AI acceptable use policy. You've got an acceptable use policy for everything else, and this is coming down the pike. So it's very important to get that out to your staff so they know — you'll have one that's designated for legal or confidential information, and then they can use non-confidential information in their own private chatbots and things like that. But to have that delineation is super important. Very good advice, appreciate that.

Jennie Woltz

I have one further thing — a really strong piece of advice that came from a recent case. There have been cases recently where people's use of AI has been discoverable in litigation, because an AI is not a lawyer. And attorney-client privilege only protects communications you make to a lawyer.

So even if you're writing in AI, "hey, I think I have a claim for X, Y, and Z, tell me, do I?" Or "do I have this workplace problem?" — right? And this can happen on the employer side, it can happen on the employee side. Those communications are not privileged.

So that's another risk that I think is important for all employers at this point in time, especially those people in HR and in legal. Be really careful about just kind of brainstorming on your own, and make sure that your communications are protected. Otherwise the whole chat history could be discoverable, and you could lose some important leverage in later litigation.

Collin McKee

Great point, thanks for bringing it up. A couple of questions before we close out. One thing that you could give as advice to any business owner this week, to do to protect themselves from everything we've talked about.

Jennie Woltz

Acceptable use policy. Make sure your staff actually knows what they're doing before using AI in their own subject matter jurisdiction. And make sure that you have the resources to get the truly specialized and expert advice when you need it.

Collin McKee

Where do you think AI and employment law is headed in the next two years, in your honest read?

Jennie Woltz

I think that AI is going to get better. It's going to be trained. I think a lot of leave administration — I do think more contracts are going to be a lot quicker. I think that policies are going to go the route of AI as well. HR — I think we're just going to see more of it.

I think one area we didn't really discuss today: employee complaints. A lot of employee complaints — we're seeing people are using AI to make their complaint seem more important than it is. So I wouldn't be surprised if we're seeing more complaints being generated from AI. But again, in the same way that there is the risk of, just because you're using AI, something might sound really good — just because an employee is using AI for their complaint doesn't mean it has any more teeth than if they were speaking it in their own words. But I do think it's going to freak people out a little bit more when they see something that looks so polished and professional, even if there's no substance under it.

Collin McKee

Use it for good, bad, and ugly. So expect it to come from all angles. And lastly, where can people find out more about you and follow your work?

Jennie Woltz

We have a website — it's wfpclaw.com. I'm also findable on LinkedIn. Those are really the two best places to find us.

Collin McKee

We'll leave links in the show notes. I really appreciate it. I think the 80/20 rule is something people can marinate on, and the sound advice for business owners to start to look at an acceptable use policy for all their employees is very valuable. I think everyone should take it back and apply it.

So if you're listening and you're wondering where your own firm or business stands in AI — whether you're exposed, or behind, or ready to move forward — that's what Endeavors AI does. We like to peel back the layers, get a good full view, and audit your business to see where things can be of benefit. And like Jennie's bringing to the fore, do it in a way that's responsible and doesn't create liability, but just creates the efficiencies that you're looking for. So reach out. Our website is endeavorsai.com.

Once again, I'm Collin McKee for Endeavors AI. Jennie Woltz, thank you for being here, and we will see you on the next episode.

Jennie Woltz

Thanks so much, Collin.