W. Crawford Appleby, a Wisner Baum partner in Los Angeles who is on the firm's AI committee, talked to Law.com about the best way to use artificial intelligence tools and how to avoid costly mistakes.
July 24, 2026 By: Amanda Bronstad
After ChatGPT became a household name about three years ago, Wisner Baum created an AI committee that reviews the latest artificial intelligence tool and trains lawyers on how to use the technology in their legal practices.
The goal is to be more efficient at work. But it’s also to avoid the traps that have ensnared other lawyers reprimanded or sanctioned by judges for AI “hallucinations,” or fake case citations, in their filings. This year, Sidley Austin apologized to a federal bankruptcy judge in New York for filing documents with incorrect case citations generated with AI. And on July 16, Chief U.S. District Judge Hala Jarbou, in Michigan’s Western District, reprimanded the Justice Department in an immigration case after one of its lawyers submitted a fake case citation, writing, “It should be obvious that any attorney who uses AI must scrupulously review its work product to ensure that the cited cases exist and that the citations accurately and fairly represent the underlying case law. The duty of candor towards this tribunal demands no less.”
W. Crawford Appleby, a Wisner Baum partner in Los Angeles who is on the AI committee, talked to Law.com about what he does and how to avoid AI mistakes. This article has been edited for length and clarity.
Q: What is Wisner Baum’s AI committee?
A: Early on, our firm decided to form a committee of folks who could be in charge of looking at this. How can this help us? What are the risks associated with this? How do we harness this technology for our work without putting ourselves at risk?
We hold regular training to teach the attorneys at the firm how to use the products and show you what it can do for you. Last week, we did a training on how to create custom ChatGPT, where you can perform certain tasks. Instead of having to do a blank slate in a new GPT chat, this is designed for a particular thing. Build a GPT and train it on how to summarize cases on this. If you want to keep track of an area of law, you find new cases, and if you want a quick summary of what the case says, you can put the case in there and it will automatically read it.
Q: Who is on the committee?
A: Four attorneys, all partners, are on the committee, and then we have some technology folks on the committee, as well.
Q: What’s been your experience with how lawyers have adopted, or avoided, AI?
A: As time has gone on, more and more folks who were hesitant to use the technology are coming around to us, seeing the value. The technology has become more effective, and tasks it can perform have grown. When people see that, they seem maybe more relevant to what they thought. But adoption with new technologies is always a challenge.
When it comes to drafting, brainstorming and trying to retrieve and collect data, I would say those are the most useful tasks for the moment. An example I personally found that I benefit from is when you have an appeal, and you’ll have oral argument, an attorney will almost always do a moot argument, where they ask their coworkers to play the role of an appellate panel and spend an hour practicing the argument with them and asking questions and think on their feet. If you could do more than one of those, it can be hard to get your coworkers or friends of yours to have the time to participate in something like that. But if, for example, you’re able to utilize AI and some of the conversation modes available, I’ve found ChatGPT to be useful. You can have AI play the role of an appellate judge and you can moot the argument all day long.
Q: What are the most common AI tools lawyers prefer to use?
A: There are a lot of different tools out there. For us, and what the AI committee’s responsible for, is evaluating new products that come out. Let’s see the demo of this, understand what this does and how it works, can it be useful? The number one thing is, how is data handled here? We don’t want our data to be used to train, we want our data closed, and don’t want to unintentionally give out data.
We have enterprise-level accounts with ChatGPT. We’re also working with Claude. Some attorneys at our firm enjoy using it for drafting and other tasks. We have a cases management system through Filevine, which has an AI component, to interact with the case files. Those are the main products we’re using. We also have enterprise-level accounts with Perplexity.
Q: What’s the biggest mistake lawyers make using AI?
A: The examples of mistakes we’re seeing everywhere are that attorneys are failing to understand the technology, and what it is. They’re trusting it too much, failing to verify the output they’re getting from it. That’s happening over and over again. It’s not designed to necessarily be accurate, but predict what the next word would be when you’re writing.
And so, when folks are asking it to be 100% accurate and know everything, they’re failing to understand what the technology is designed for. At times, depending on how it is used, it can be a square peg in a round hole. You have to understand what it is and how to use it properly. And even if you do that, you have to verify that. The current cure for this is humanity and human verification. Is this really going to save me time if you verify it? Let’s say it took you a half hour to write something, and AI can draft the same thing in a couple of minutes. If you take five minutes to review that and make sure it’s accurate, the cases are true, and say what the AI has written, you’ve saved yourself 20 minutes.
Q: Judges across the country have reprimanded, or sanctioned, lawyers whose use of AI has created fake “hallucinations,” or incorrect case citations. And they aren’t just small firms; even the DOJ and Sullivan & Cromwell got disciplined. What are law firms doing wrong?
A: What’s going in is AI creates the temptation to cut corners because it creates confident-sounding answers. And the words are very impressive. So, when you combine that with how much pressure attorneys are under to produce quality work, there are situations where lawyers fail to verify the output from AI and there’s a fake case citation in a court document.
Q: It’s not just judges. What are the risks that clients will file malpractice lawsuits if lawyers use AI that makes errors?
A: There’s an element that requires the attorney, in this day and age, to be aware of new technology, and a level of requirement there. It could turn into an argument that their conduct as the attorney fell below the standard of care. There is a potential for clients to bring malpractice claims against attorneys for misusing the technology based on ethical rules.
I would point to the one that seems to be the most prevalent in hallucination cases, and the recent DOJ case, in which the judge cited the attorney’s duty of candor. American Bar Association Model Rule 3.3: It says attorneys are not allowed to submit false statements under doctrine of law to the court. When you file a brief with a case in it that’s not real, you step into that zone.
Q: How can law firms ensure that both their court filings, and the legal research, remain accurate when using AI?
A: When it comes to AI adoption by law firms, it’s best to take a slow and steady approach to that. Don’t rush into the field. It’s exciting to see a story or headline: "We’ve adopted AI in our practice and look at how amazing we’re doing!" That’s great. But it’s important for law firms to take a slow and steady approach. You don’t want to be the hare in AI; that’s how you end up being in the headlines. You want to be the tortoise.
Source: Law.com
Published: July 24, 2026
