Artificial intelligence may be ushering in welcome changes to the mortgage industry, but for servicers in particular, it's also fueling a more troubling trend: a rush of borrower lawsuits filed with the help of AI tools like ChatGPT.
In a trend appearing with greater frequency throughout corporate law, mortgage servicers find themselves confronting a significant rise in the number of suits filed against them by plaintiffs choosing to represent themselves typically when facing the threat of foreclosure.
"2024 really marks the beginning point,"
While no clear evidence might show up at first glance unless litigants willingly attest to using the technology, AI's fingerprints are discernible all over cases, at times literally in response to defense counsel's arguments.
"When we file a motion to dismiss, telling the court that they should dismiss the borrower's claims against our client because they don't meet the applicable legal standard, we'll have a 25-page response filed to that motion within 20 minutes," Kolodziej said. "That's obviously an extreme example, but those extreme examples happen, and that's a dead giveaway."
Although serving as one's own lawyer in court, referred to as pro se representation in legal circles, is a long-established, if infrequent, practice, the arrival of AI and its rapid improvement is leading to a noticeable surge in such cases. Servicers can expect to see heavier workloads and changes in legal strategy to help address the jump in pro se cases brought forth by distressed homeowners this decade, attorneys say.
Between fiscal years 2019 and 2023, the total number of non-prisoner pro se civil lawsuits filed in U.S. federal courts accelerated by over 52% from 25,846 to 39,408. Corresponding with the emergence and spread of the initial artificial intelligence boom, pro se filings jumped another 39% on a yearly basis by the end of FY 2024 to 54,675, according to the Administrative Office of the U.S. Courts.
Numbers settled back downward in 2025, dropping to 41,788, with the total still well above the 2019 mark.
"A determined pro se who's dedicated to the cause can be time consuming and difficult," said Loren Coe, partner at Stinson. "There may be no merit to legal arguments being made, but the volume that can be generated in motion practice and correspondence and disputes and everything else can be frustratingly overwhelming."
Reasons behind the rise
Lawyers and the companies they represent are grappling with a technology that can generate highly flawed output but still manages to come off as convincing.
"One of the more dangerous parts of AI is that it is always very confident whether it's right or not. It's not good about saying, 'I don't know the answer to this,'" said Cliff McKinney, managing member at Quattlebaum, Grooms & Tull and an adjunct professor at Bowen School of Law at the University of Arkansas at Little Rock. McKinney is also the
Lawyers agree that on the surface, AI-generated legal filings look professional and legitimate. Yet numerous examples of judges finding hallucinated citations across various types of cases have already been reported, leading to admissions of AI-generated pleadings by lawyers who were subsequently penalized.
"It's a lot easier for courts to say a claim has no validity when it's chicken scratch, but now, when you have something that seems plausible, and they're citing prior precedent or supposed prior precedent, that changes how we have to handle it," Kolodziej said.
While the technology has the ability to whip out a file within seconds, the growth of self-representation in the courts can also be attributed to other factors, such as limited availability of resources for mortgage and consumer delinquency cases, all of which makes AI "an easy tool to jump on," said McKinney, who has experience assisting pro se litigants in access to justice programs.
Likewise, changes in servicing rules, which shifted the
"A lot of folks are just desperate. They can't afford to hire an attorney, and they're looking for somebody to help them. They end up doing it on their own," McKinney said.
What servicing attorneys see today
Beyond obvious examples where fabricated citations appear, experienced lawyers now find telltale signs of AI-generated content beneath the hood of pro se upon closer examination. Their findings underscore the heightened scrutiny now required from legal counsel when it comes time to mount a defense, they say.
"The hallucinations are less common. But now what it claims a case is saying is not always 100% the way we would interpret it," Kolodziej said. "Cases holding one thing may not be entirely applicable to somebody else's circumstances."
AI also reveals itself through the near-identical structure and presentation of arguments across cases it writes, Coe said. The claims presented often fail to hold water due to the technology's habit of drawing upon inaccurate or dated source material. "These theories have no merit whatsoever, but there's a resurgence all of a sudden," Coe said.
Borrowers are filing such cases after asking AI to mine the internet for winnable theories or examples of past foreclosure claims and solutions, he surmised.
"They're retrieving a bunch of old theories that, honestly, we haven't seen much of in the last decade. It's sovereign-citizen-type arguments or separation of the note from the deed of trust — a lot of things we saw in 2010 through 2014 that have been thoroughly debunked by state and federal courts."
Yet even flawed AI drafts can't be discarded or ignored by defense attorneys, with substantially more effort now needed to properly review and verify claims and citations they previously could assume were correct.
"It's not just looking to see, 'is this case real, but does it stand for the proposition that the person claims it does? Are the quotes in it real?' I've seen some instances where AI has given me a statement that is completely false, but they've cited a completely true case," McKinney said.
The investment of additional personnel and effort means servicers and others fighting this litigation should expect to see heftier legal bills.
"Our clients are understanding of the fact that sometimes these pro se cases actually cost a lot more money to defend," said Minh Vu, partner at Seyfarth Shaw.
Examples of such litigation extend
The use of AI by some pro se plaintiffs knowledgeable about legal protocols can be "problematic" because they understand how to work the system to their advantage with no fear of reputation or career risk unlike her professional peers, she continued.
"When you have pro se plaintiffs, they don't have the threat of being disbarred hanging over their head. They don't care. There's nothing at stake."
Instead, they may view their lawsuits as a strategy to delay undesirable outcomes or perhaps helping them reach a settlement with defendants, attorneys say.
Outside of the courtroom, AI has also gained traction among borrowers who use it to submit complaints to state and federal regulatory agencies, such as the Consumer Financial Protection Bureau. Even if their allegations turn out to be baseless, companies at the center of the inquiries are obligated to respond, increasing the workload primarily for in-house counsel.
"Whether or not they're ultimately determined to be meritorious, the servicers do a good job of looking at everything, investigating, reviewing, providing responses. It's just time consuming," Coe said.
How the legal community could respond to AI use
With artificial intelligence now both used and misused in legal cases, attorneys hope guidance addressing what appropriate AI usage in the courtroom will emerge, at the same time they welcome the benefits it brings to their profession.
Requiring litigants to adhere to existing rules would go a long way to limiting some of the more baseless lawsuits that eat up legal departments' time and money, according to Vu. In particular, the rules requiring attestation that the contents of pleadings are based on facts and thorough review could be better enforced by judges, she continued.
"Courts have to be more vigilant about applying them and calling out and sanctioning behavior that is inconsistent with those rules. Unfortunately, not that many courts aggressively manage their cases," Vu said.
The growth of AI in judicial matters also can lead to the creation of systems that ensure counselors, including pro se litigants, understand the risks of using the technology, the signs of its application and proper and honest usage, according to McKinney.
"I think it could help people who maybe can't afford a lawyer eventually. It can also hurt, too. It's a tool that has to be used responsibly," he said.
Judges, including one adjudicating a recent servicer case in California, are beginning to heed some of those calls. Several federal district court judges in the past year have issued standing orders mandating counsel not only to sign attestations if AI was used in drafting documents, but also to describe how and vouch for its accuracy.
The orders should serve as a deterrent that can limit the spread of frivolous lawsuits, according to Kolodziej. At the same time, AI has already made pro se litigation a fast and straightforward process, meaning volumes are likely to remain elevated for the time being in the absence of widespread accepted guardrails.
While AI may be creating more case volume, it won't necessarily translate to better legal arguments, he added.
"I don't think the outcomes are really changing all that much," Kolodziej said. "Our clients, way more often than not, are complying with their obligations, and they're just having to fend off oftentimes meritless lawsuits."
