Courts across the country are cracking down on AI-drafted filings and contracts. And the person paying the price? Those who use AI only.
It starts innocently. You’ve booked an attorney through HelloPrenup, and while you wait on their review, you paste a clause into a chatbot “just to double-check.” It’s free, instant, and endlessly confident. Within minutes, you have a list of red flags, revisions, and valid arguments, and your attorney’s advice starts to look like just one opinion among many.
The problem: chatbots are wired to please you. They want to give you something. And sometimes they are wrong in ways a non-lawyer can’t see. The consequences? They’re showing up in court. Judges are receiving AI-drafted filings and contracts that cite cases that don’t exist, misstate the law, or argue positions no competent lawyer would take. Judges are responding with sanctions, dismissals, and new rules that make clear the courtroom is not a place to experiment with AI.
These same habits that sink a court filing can also quietly weaken a prenup.
Family law shows the trend most clearly. In a spring 2026 survey of 133 lawyers at the national family-law firm Cordell & Cordell, 80% said clients had sent them correspondence, draft pleadings or strategy ideas they believed were AI-generated, The Wall Street Journal reported.
What’s happening in courtrooms
Since 2023, judges have been fighting against AI mistakes. Here’s what a judge can do when they see improperly used AI:
Sanctions for fake citations. The case that put this on the map was Mata v. Avianca (S.D.N.Y. 2023), where attorneys were fined $5,000 after submitting a brief citing cases ChatGPT invented. Since then, courts have sanctioned lawyers at small practices and large national firms alike, in some cases removing them from the matter or referring them to bar regulators.
Filings struck or dismissed. When a filing is built on nonexistent authority, judges increasingly refuse to consider it. In Kruse v. Karlen (Mo. Ct. App. 2024), a self-represented appellant’s brief contained roughly two dozen citations, nearly all fabricated. The court dismissed the appeal and ordered him to pay $10,000 for filing a frivolous appeal.
New standing orders and local rules. Many judges now require parties to certify whether generative AI was used in a filing and that a human verified every citation. Judge Brantley Starr of the Northern District of Texas issued one of the first such orders in 2023; many courts have followed with their own versions.
Keep in mind: self-represented people still have a legal right to file on their own behalf, and no court has banned them from doing so. But courts are drawing a much harder line on what they will accept. Repeat filers of baseless or fabricated motions can face pre-filing restrictions that require a judge’s permission, or an attorney’s signature, before anything new is docketed.
In practice, a lawyer has become the surest way to get a motion taken seriously.
That line is already being drawn. A Florida husband representing himself in his divorce, using AI, filed at least 145 motions and then sent an appeals court 45 AI-generated pages plus an 863-page appendix. In May 2026, the court ordered its clerk to reject his future filings unless a lawyer signs them, according to the WSJ.
Why people without lawyers are hit hardest
Trackers of AI-hallucination cases in court have logged hundreds of incidents worldwide, and a large share involve self-represented litigants. That isn’t because they’re careless. It’s because they’re relying on AI and they have no way to catch the errors.
Damien Charlotin, an adjunct professor at UC Law San Francisco who tracks these errors, has counted 148 family-law rulings alone in which judges flagged fake citations or other apparent AI mistakes, with more than 100 of them in 2026, per the WSJ.
A trained lawyer reading an AI draft notices when a case name looks off, when a quoted holding sounds too convenient, or when an argument ignores the procedural rules of that particular court. A non-lawyer sees a document that looks exactly like what a lawyer would produce, and has every reason to trust it.
The consequences also land harder. A sanctioned attorney has insurance, a firm, and experience negotiating with the court. A self-represented person can lose the case outright, owe the other side’s legal fees, and damage their credibility with the judge for the rest of the matter.
It also affects the other side. A Denver family lawyer, Maha Kamal, described a self-represented father who flooded the court with nine apparently AI-written filings, one citing a legal principle that didn’t exist. The judge warned him he could be ordered to pay the mother’s legal fees. “It could be used as a form of litigation abuse,” Kamal told the WSJ.
How AI talks you into bad legal ideas
AI doesn’t set out to mislead anyone. But the way these tools work makes them very good at convincing people of things that maybe aren’t true, which can feel a lot like being gaslit.
- AI invents authority: Language models generate text that sounds right, not text that is verified. A made-up case with a realistic name, court, and year is easy for the model to produce and hard for a non-lawyer to spot.
- AI agrees with you: Chatbots tend to follow the user’s framing. Ask “Are there any issues with this prenup?” and you’ll get an argument for why there are issues, not an honest assessment of whether this prenup would hold up in court or be beneficial to the user.
- AI never sounds unsure: A wrong answer arrives in the same calm, authoritative tone as a right one. There’s no hesitation, no “I’d want to check that,” no warning that this court has a different rule.
- AI doesn’t know your judge: Deadlines, local rules, filing formats, and which judge you drew all shape what a winning motion looks like. Yes, even judge by judge can vary the process. A general-purpose tool has no reliable view of any of that.
Two cases reported by the WSJ show how this plays out:
- He had to apologize to the judge: Logan Burns, a Georgia father in a Montana custody fight, skipped a $7,000 lawyer and paid $19.99 a month for ChatGPT instead. It pointed him to the six-month residency rule, so he wrote the judge arguing the court lacked jurisdiction. The judge sternly corrected him: the rule has exceptions for emergencies, and Burns had filed an emergency request himself. “It doesn’t give you the interpretation of law,” Burns later admitted.
- Convincing your lawyer to believe the AI won’t work either: A business owner asked a chatbot whether a judge would cut his spousal support after his income fell. It told him he had an easy win, without ever asking about his financial statements or spending. Trusting the chatbot, he hired a lawyer and talked the attorney into following its strategy. In late 2024, the lawyer asked the court to reduce his support and attached company records to back it up. The other side’s lawyer used those same records to argue his income hadn’t really fallen, and about a year later the judge agreed. He blamed his lawyer, not the chatbot, and moved to a new firm, where attorney Hossein Berenji said, “It took us a while to convince him to move away from what AI was advising him to do.” That firm brought in a forensic accountant, who found what the first filing missed: spoiled inventory had eaten the profits even while sales stayed high. What the chatbot called an easy win took two law firms and nearly two years.
Put together, these traits can walk someone step by step into a confident, well-formatted filing or contract that hurts their own case.
What this means for your prenup
A prenup is only worth something if a court will enforce it years from now and the terms inside of it are beneficial to you now, and also in the future. AI doesn’t always know how to do this.
- Enforceability depends on your state, and some rules are hard to interpret: Prenup rules vary by state, including what must be disclosed, how far ahead of the wedding it should be signed, and which terms a court will refuse to enforce. Some rules are unwritten rules buried in case law that aren’t available to the general public, too. A chatbot’s answer can’t always reliably tell you what holds up where you live.
- Some terms don’t belong in a prenup at all: Courts generally won’t enforce clauses that decide child custody or limit child support in advance. An AI-suggested clause like that doesn’t just fail; it can give someone a reason to challenge the rest of the agreement.
- One-sided terms can backfire: A chatbot will happily help you push for everything. But an agreement that looks grossly unfair, or that your partner signed without a real chance to review, is easier to attack later.
- Your partner has an attorney too: AI-generated redlines tend to be long and adversarial. Sending them across the table can stall negotiations and strain the relationship the prenup is meant to protect.
- It costs you more: Every chatbot revision your attorney has to review and rebut is time you pay for, without making your agreement any stronger.
What an attorney gives you that AI can’t
The value of a lawyer was never just knowing where to find the law. It’s judgment, accountability, and a relationship with the court.
AI can even make good representation more expensive. Toronto family lawyer Steven Benmor told the WSJ a client sent him a chatbot-driven, 13-part list of revisions nearly as long as the prenup he’d drafted. His advice: “Don’t spend money unduly by asking a human expert to enter into a debate with a robot expert.”
- Verification: An attorney reads the actual cases, confirms they say what the brief claims, and checks they’re still good law. What is still considered “Good Law” is often hidden behind legal database paywalls that many chatbots don’t have access to.
- Honest advice: A good lawyer will tell you when your argument is weak, when to settle, or when not to file at all. That “no” can be worth more than any motion.
- Strategy: Knowing which motion to file, when, and in front of which judge is learned from experience, not text prediction.
- Accountability: Lawyers sign their filings, answer to the court and the bar, and carry malpractice insurance. If something goes wrong, there’s someone responsible.
- Credibility: Judges and opposing counsel treat a filing differently when a licensed professional stands behind it.
As courts raise the bar for what they’ll accept, these qualities are moving from helpful to essential.
Using AI wisely
AI isn’t the enemy. It can help you understand legal vocabulary, fill in knowledge gaps, organize your timeline of events, or prepare questions before a consultation. Used that way, it can make your time with an attorney more productive and less expensive.
Where it becomes dangerous is when it replaces professional judgment: drafting clauses, redlining your attorney’s work, or deciding what’s fair or enforceable.
If a chatbot is telling you exactly what you want to hear, that’s likely a problem.
The courts have made their position clear: the tools are new, but the standard isn’t. What you file, and what you sign, has to be accurate and grounded in real law.
If you’ve booked an attorney through HelloPrenup, let them do the job you hired them for. Bring every question you have, including ones a chatbot raised, but treat AI’s answers as questions to ask, not conclusions to argue. A prenup built on your attorney’s judgment is the one most likely to hold up when it matters.
Julia Rodgers is HelloPrenup’s CEO and Founder. She is a Massachusetts family law attorney and true believer in the value of prenuptial agreements. HelloPrenup was created with the goal of automating the prenup process, making it more collaborative, time efficient and cost effective. Julia believes that a healthy marriage is one in which couples can openly communicate about finances and life goals. You can read more about us here Questions? Reach out to Julia directly at Hello@Helloprenup.com.
0 Comments