Self-representation is not a fringe phenomenon. By one frequently cited estimate, roughly one in six Americans appears in court each year without a lawyer, and in civil cases the share is far higher, with studies suggesting that about three in five civil litigants proceed without counsel. The numbers are climbing rather than holding steady. A New York State Bar Association article reports that pro se litigants in that state's unified court system numbered around 83,000 by the end of 2024, and that self-represented parties accounted for 28% of all federal court cases between 1999 and 2018. Courts have long regarded this population as a strain, because pro se litigants tend to consume more of a judge's and clerk's time and to struggle with the formalities that lawyers handle as a matter of course.
For most of that history, one thing kept the strain manageable: drafting was difficult. A would-be litigant had to find a cause of action, line up the elements, get through service, and produce something a judge would not dismiss on sight. The effort never screened out bad claims, but it rationed them, holding back the person inclined to sue repeatedly. That barrier is now gone. Generative artificial intelligence produces a passable-looking complaint, motion, or brief in about the time it takes to describe the underlying grievance. It does not tire, it does not bill by the hour, and it does not lose interest after the fifth or sixth filing, the way a determined pro se litigant once did. A single self-represented party can now generate the paper output of a multi-lawyer firm, and some are doing exactly that. The vexatious-litigant problem that Florida's courts have generally wrestled with for decades, and that the Legislature was slow to address, has not merely persisted. It is accelerating.
The problem is no longer hypothetical
Consider Avery v. Beauzil, 2026 WL 1475262 (Fla. 4th DCA May 27, 2026). The husband, in a divorce case, filed a 41-page, single-spaced mandamus and prohibition petition that the Fourth District found to be AI-generated, then supplemented it with 24 and 19 additional single-spaced pages of AI-generated argument, and more than 1,000 additional AI-generated documents and charts. The petition sought rulings on more than 90 pro se motions he had filed below; he later acknowledged having filed at least 145. When the court ordered him to show cause why it should not stop accepting his pro se filings, he answered with a forty-five-page AI-generated response and an eight-hundred-sixty-three-page appendix of still more AI argument. The court barred his pro se filings, directing that nothing further be accepted unless signed by a member of the Florida Bar.
Avery is a vivid illustration of the volume problem, but it also directly names the reliability problem. The petition argued that orders entered by a disqualified judge were void and cited real decisions for that proposition; the court found the argument misstated the law and that the cited cases did not support it. Such AI-hallucinated arguments, the court held, are just as egregious as a citation to non-existent law, and pro se litigants are not held to a lesser standard for them. A footnote captures the judicial mood: the court had seen this hallucinated argument often, and seeing it at all was too often.
The fabricated-authority problem appears on its own in Gleason v. Marcus, 2026 WL 1311959 (Fla. 2d DCA May 13, 2026), where the Second District affirmed a dismissal but felt compelled to address the litigant's apparent use of AI: his briefs cited numerous cases that do not exist. They quoted real decisions for propositions that were nowhere to be found in them. The court's borrowed formulation is the heart of the matter: the technology is very good at sounding right but less adept at being right, and blind reliance on it is unwise, unacceptable, and sanctionable. The court tied the warning directly to the signer's accuracy obligation under Florida Rule of General Practice and Judicial Administration 2.515(d)(2).
The most thoughtful treatment comes from Gouveia v. Meridian Financial Investments, LLC, 2026 WL 816497 (Fla. 4th DCA Mar. 25, 2026), where the briefs were replete with citations that either did not exist or addressed unrelated issues, one supposed Florida case in fact leading the reader to a Mississippi punitive-damages decision. Judge Lott, in his concurring opinion, reasoned that the attorney AI problem is largely self-correcting because attorneys are repeat players who learn from sanctions, can pay them, and face discipline if they persist. Pro se litigants are different. They are usually one-shot players who never see the warnings issued in other cases, who do not appreciate why something that sounds right is not right, and who, because the cost of generating content is so low, can produce a great deal of it. The result, in Judge Lott’s words, is a seemingly endless deluge that forces opponents and courts to spend more resources disproving AI output than parsing an ordinary pro se appeal. Sanctions imposed on a case-by-case basis do nothing to deter the next unwarned litigant; the real task is a front-end solution.
This problem is not without tension. There is a shortage of affordable legal representation, and for many people without counsel, an AI chatbot is the first accessible substitute they have ever had. One legal-aid leader quoted in the New York State Bar's coverage compared the technology's potential, in processes built on forms and letters, to the invention of fire. That promise is not imaginary. The same technology that lets a vexatious litigant fire off ten near-identical complaints also lets a wronged person who cannot afford a lawyer state a real claim she could not have drafted on her own. Inexpensive drafting is not all bad; it widens the courthouse door for the meritorious and the meritless at once. The Avery petitioner pressed that point, arguing that as a disabled veteran, he had a right to use AI because counsel would cost him far more than the twenty dollars a month he paid for his AI service. The court rejected the notion that this licensed an abuse of process, and the distinction it drew runs through all of this. The law does not sort filings by who or what drafted them. It sorts them by conduct, accuracy, and result.
The statute built for the pattern: Florida Statute Section 68.093
If the conduct is a pattern of meritless filing, Florida has a statute aimed squarely at it, one the Legislature left substantially untouched for a generation before amending it in 2025. Florida's Vexatious Litigant Law, codified at section 68.093, Florida Statutes, rests on a straightforward premise: a litigant who repeatedly files meritless actions imposes real costs on the courts and the parties forced to defend, and at some threshold that pattern justifies restricting his otherwise broad access to the courthouse.
The statute works in two stages.
First, it defines who qualifies as a vexatious litigant. As amended in 2025, a vexatious litigant is a person proceeding pro se who, in the immediately preceding seven-year period, has commenced, prosecuted, or maintained five or more actions that were finally and adversely determined against him, unless the court finds a particular action was brought in good faith. The revision extended the lookback period from 5 years to 7 years, widening the pool of eligible filers. It also reached beyond whole lawsuits to the litigant who repeatedly relitigates settled issues, files unmeritorious pleadings, conducts unnecessary discovery, or uses tactics that are frivolous or intended solely to delay. And it extended the law to actions under the Family Law, Probate, and Small Claims Rules, which the original version had largely left out. An action does not count as finally and adversely determined while an appeal in it is pending.
Second, the statute supplies two distinct remedies. Subsection (4) authorizes a prefiling order: a court may bar a vexatious litigant from commencing, pro se, any new action in that circuit without leave of the administrative judge. That is prospective; it governs the next lawsuit, not the one on the docket. Subsection (3) authorizes a security requirement. On a noticed motion and after a hearing, a court may require a litigant shown to be vexatious and not reasonably likely to prevail to post security covering the moving party's anticipated fees and costs. If the plaintiff fails to post it, the court must dismiss the action with prejudice as to the moving party. Dismissal of the pending case, in other words, runs through the security mechanism of subsection (3), not the prefiling injunction of subsection (4). The 2025 amendment added teeth as well, allowing clerks to refuse filings from a litigant under a prefiling order, providing that an automatic stay lasts until the court acts, and confirming that the statute is cumulative with other remedies, including section 57.105 safe-harbor sanctions.
With the mechanism laid out, the recent decisions show courts working through its finer distinctions, with the appellate courts settling the boundaries the text leaves unclear.
Courts working out the statute's reach
The clearest illustration of courts working out the statute's reach is the line between its two principal remedies. In Mistivar v. Broward County Sheriff's Office, 427 So. 3d 45 (Fla. 4th DCA 2025), a trial court declared the plaintiff vexatious and, under subsection (4), both barred his future pro se filings and dismissed his pending case. The Fourth District affirmed the vexatious designation but reversed the dismissal, holding that subsection (4) applies only to future actions, not to the one already on the docket; dismissal of a pending case instead runs through the security mechanism of subsection (3). The companion order on remand, Mistivar v. Broward County Sheriff's Office (BSO) (Fla. 17th Cir. Ct. Jan. 29, 2026), set out the proper course, dismissing only after the litigant was ordered to post security and failed to do so.
A finer line separates the statute's own remedies from a court's inherent authority, and Steinberg ex rel. Fiesta Homeowners Association, Inc. v. Cudak, 429 So. 3d 21 (Fla. 4th DCA 2026), shows the courts policing it. A trial court there declared a pro se homeowner vexatious, imposed a security bond, and directed the clerk to reject his future filings unless signed by a lawyer. The Fourth District upheld the vexatious finding but reversed the filing bar because barring pro se filings under a court's inherent authority requires an order to show cause and a chance to respond, neither of which the litigant received. The same case confirmed that a litigant cannot evade adjudication by dismissing his case a step before the order.
The newest decision Walters v. State, 51 Fla. L. Weekly D1233a (Fla. 3d DCA June 17, 2026), affirmed the order requiring the posting of a $10,000 security bond to continue to prosecute a lawsuit by a pro se plaintiff who had filed forty-five losing actions in seven years.
The accuracy backstop: Rule 2.515
If section 68.093 screens the litigant, the Florida Supreme Court's newest measure screens the document. On May 28, 2026, on its own motion, the Court amended Florida Rule of General Practice and Judicial Administration 2.515(d)(2) to require that the signer of any filing represent that the legal authorities identified in it exist and are accurately cited. See In re Amendments to Florida Rule of General Practice and Judicial Administration 2.515, No. SC2026-0673 (Fla. May 28, 2026). The amendment took effect June 15, 2026, and applies equally to filings by attorneys and by unrepresented parties. It carries an express sanctions provision: a court may, on its own motion or a party's, sanction any filing inconsistent with the representation after notice and a hearing, with sanctions ranging from reprimand and contempt through striking the document, dismissal, and an award of fees and costs.
The Court was candid about why. Generative AI tools, it noted, have become routine drafting and research aids, and for all their usefulness, they also produce content that reads as plausible while being flatly wrong, hallucinated authorities among the worst of it. The amendment requires no separate AI-disclosure form and no certification that AI was used. It shifts the question from how a filing was produced to whether its citations are real, which is the question that matters. A lawyer who hand-types a fake citation is in no better position than one who pasted it from a chatbot. Signing is the representation, and the duty to check falls on whoever signs, regardless of which tool produced the draft.
What the courts now face
Put the pieces together, and the shape is clear. Section 68.093 reaches the litigant whose history of meritless suits has earned a limit on his access; Rule 2.515 reaches the filing whose authorities are fabricated. One screens the person, the other the document, and neither asks who did the drafting. For a quarter century, the statute was a last resort against the occasional obsessive whose persistence outran the system's natural friction. AI removes the friction. It hands a single pro se litigant the output of a firm and, at the same time, corrupts the reliability of that output, so the courts now face more filings and less trustworthy filings at once, from litigants newly able to litigate vexatiously at scale.
The open question is whether the courts can keep pace. The remedies are individualized: each requires a motion, a showing, and, for security, a hearing. Enforcement of Rule 2.515 depends on someone catching the fabricated authority and running it through notice and a hearing. As Judge Lott observed, sanctioning one litigant after the fact does nothing to deter the next who never saw the warning. If meritless and inaccurate filings rise faster than courts can handle them one at a time, the framework will work beautifully in any single case but poorly in the aggregate.
The answer is not to abandon these tools but to reach for them earlier and more routinely than the bar has been accustomed to, and to take the front-end fixes, disclosure, and certification seriously at the point of filing, that judges confronting the deluge are beginning to call for. The old drafting barrier is not coming back. Section 68.093 and Rule 2.515 are what replace it, and for litigators facing a pattern of meritless pro se suits, for anyone who now vouches for every citation she files, and for courts working dockets that cheap drafting will only swell, they have stopped being curiosities at the margin. They belong near the middle of civil practice now.
This article is offered for general informational purposes only and does not constitute legal advice. It does not create an attorney-client relationship, and it should not be relied upon as a substitute for the advice of qualified counsel regarding the specific facts and circumstances of any particular matter.