"There was not much in there, and we thought it was a weak case," said Long, a partner at Nelson Mullins Riley & Scarborough LLP.
But once Long responded to the complaint, he and his client were hit with a flurry of motions — 21 in total — sometimes coming at a clip of nearly one per day.
The pro se plaintiff was using artificial intelligence to draft his barrage of challenges, something that employment attorneys say is becoming increasingly more common, with the potential to clog up dockets, drag out cases, and make litigation significantly more expensive.
"We are absolutely seeing an uptick in AI use by pro se litigants," said Eric Felsberg, the leader of Jackson Lewis PC's subgroups for AI governance and bias testing and pre-employment assessments.
"I won't say it's easy to spot, but there are definitely signs you can pick up on," Felsberg said. "You'll see very polished submissions. The papers will bring up good, solid, legal arguments that pick up on complexities that you would not expect to see in a pro se submission."
While AI has been touted as a way to provide access to the courts for individuals who might otherwise be mystified by navigating the legal system without a lawyer, not all attorneys welcome the trend.
"Not only are there more lawsuits, but they are harder to kick," said Sal Simao, another Nelson Mullins employment attorney who practices in New York and New Jersey.
"Are they winning lawsuits? Probably not. But we are seeing people file appeals to the Third Circuit because the AI tells them how to do it," Simao said. "It's becoming harder and harder to budget, because you don't know what's going to happen. It's not a normal case."
"He Kept Us on Our Heels"
For weeks, Long and his team navigated the onslaught of motions, which included a motion to strike their answer to the complaint, a motion to prevent them from being admitted pro hac vice in Virginia, and motions for reconsideration if the first attempt were denied by the judge.
"I'll be honest, they were well-written," Long said. "They were presumably citing good law, and we had to respond to them day after day after day after day. The guy kept us on our heels for the first three months of the case."
Howie Wexler, a labor and employment lawyer with Seyfarth Shaw LLP, said he first encountered a pro se plaintiff using AI last year when helping a client try to resolve a charge with the U.S. Equal Employment Opportunity Commission. Since then, he said, the use of AI by pro se plaintiffs has "increased tenfold."
Felsberg said the use of AI is "empowering pro se plaintiffs" and can sometimes be helpful to defense attorneys because filings are easier to understand than they would be in most pro se matters.
"Before, if you got a complaint, it may be, for lack of a better description, kind of a rambling statement," he said. "We'd have to ask: 'Is a claim even asserted here? What violation of the law are they claiming?'"
But while AI can streamline a pleading and make it easier to read for an attorney, it isn't always helpful, even for the plaintiffs using it. Some of the well-worn AI pitfalls like hallucinations can permeate filings, a trap experienced attorneys have occasionally fallen into in various types of cases.
Wexler stressed the need to check every single case citation in pro se matters to make sure the cases being relied upon are indeed real.
"That was not something that was a concern several years ago," he said. "But I have seen it more in the past six months than I had in 20 years of practicing law."
Kellen Safreed, a plaintiff-side employment attorney with Sherin and Lodgen in Boston, said he often sees potential clients come in with AI-aided pleadings information that is "true, but irrelevant."
"They will bring a summary that includes citations or statutes or cases out of Indiana," Safreed said. "Maybe that's a real thing, but it has nothing to do with their Massachusetts employment claim."
AI also has a tendency to tell the user what the user wants to hear, Safreed added.
"Someone wants to quit their job, but they want to get severance, and AI says, 'Sure, it's called constructive discharge,''' he said. "They get their hopes up, but they don't realize that is a really high bar. Your boss being miserable to the point where you want to quit your job is probably not going to be a constructive discharge claim."
Simao said AI is making it harder to know whether a document is real, citing a recent case in which a pro se plaintiff created a picture and audio recording to support their case and a notarized document that his Nelson Mullins team thought was real.
"We took the extra step of calling the notary, who couldn't remember but checked her log and then told us it was a forged document," he said. "While authenticating evidence is always an issue, pro se plaintiffs tend to be treated more leniently by courts, and forgeries are getting harder to detect."
ChatGPT-Client Privilege?
The use of AI can also invite complicated questions around privilege. Safreed said there is "huge uncertainty" about where and when attorney-client privilege would apply when AI is involved.
"If a client uses AI to summarize their claims, is the record that exists of that exchange privileged? I don't know," he said. "I think that's an open question."
A company defending a claim could potentially file a motion seeking to compel production of the ChatGPT prompts from a server, he said.
"I think employees should be cautious about that," Safreed added.
Seyfarth's Wexler envisioned the possibility of a whole new category of evidence that litigants are going to want to get their hands on, related to communications between a person and a public AI platform.
"There is a potential treasure trove of information," he said.
In Long's Virginia case, he said that the plaintiff — who made no secret of the fact he was using AI to help advance his case — may have repeatedly entered different potential claims into ChatGPT to see which one had the best chance to stick. His colleague, Simao, said that is common among pro se plaintiffs.
"Basically, if they went to an attorney, the attorney would say, 'You don't have a claim,'" Simao said. "If you chat on one of the OpenAI systems and say you are having an issue at work, you can ask, 'What can I sue for?' The AI will make a claim for them. If they don't have the facts, they will get the facts."
Experts said it's important for attorneys to protect their own privileged information while dealing with a pro se plaintiff using AI. Long advised trying to obtain a protective order early on in the case to bar the plaintiff from running a client's documents through an AI platform, something he said happened in the Virginia lawsuit.
A Loss for Everyone
Ultimately, when it came time to deal with matters in person, Long's experience won the day. The plaintiff showed up for a deposition with a friend of his toting an old-school video camera and tried to record the deposition without a court reporter, something that was not allowed in the southern Virginia state court and was shut down by the judge.
By the time the plaintiff was able to get a court reporter, the discovery clock had run out. Long said the plaintiff struggled during a dispositive motion hearing, even though his written responses to their filings had been fairly strong.
"ChatGPT couldn't help him at the hearing," Long said.
Long, who gave a seminar to his colleagues detailing his experience, said the use of AI in this case was probably bad for everyone. Although he won a summary judgment motion for his client, the case was much longer and more expensive than it otherwise would have been. Long also said AI made the plaintiff believe his case was stronger than it actually was, which prevented a settlement.
"We made some very reasonable offers, and he just turned his nose up and said 'no,''' Long said.
He said there needs to be additional rulemaking or possibly legislation governing the use of AI in legal filings. He also predicted that judges, who are typically lenient when it comes to pro se plaintiffs, might become stricter if their dockets are jammed up by meritless cases that drag out through extensive motion practice.
"If you have a meritorious claim, AI serves a good purpose for those who cannot afford counsel, and lawyers are not getting cheaper," Long said. "But it also empowers those with frivolous cases to take them further than they thought. Unchecked, it's going to create serious issues for our courts."
Safreed said he could understand why individuals would turn to AI, despite its pitfalls, if they felt they had a lawsuit to bring.
"A company that gets sued all the time, they know exactly what to do. But I deal largely with people who have never had to engage an attorney before and, hopefully, never will again," he said. "Often, they are perfectly reasonable people who are unfamiliar with the legal machinery. That invites the usage of AI."
--Editing by Robert Rudinger.
For a reprint of this article, please contact reprints@law360.com.
