The Numbers Are Not Small

Damien Charlotin, a researcher at HEC Paris business school, tracks every court sanction worldwide tied to AI-generated errors in legal filings. His count as of early 2026: more than 1,200 total, with roughly 800 from U.S. courts alone.

Ten cases from ten different courts hit in a single day recently. The rate started accelerating in 2024 and has not slowed down. Early public embarrassments — the kind that were supposed to scare people straight — apparently did not.

Lawyers Are Not Off the Hook

This is not just a story about amateurs misusing tech. Licensed, credentialed attorneys are doing this too.

The lawyers representing MyPillow CEO Mike Lindell were fined $3,000 each for submitting briefs loaded with fictitious AI-generated citations, according to NPR. That made headlines. Other lawyers kept doing the same thing.

Last month, a federal court in Oregon ordered a lawyer to pay $109,700 in sanctions and costs for filing AI-generated errors. That may be a new record, according to Charlotin.

In February, the Nebraska Supreme Court grilled Omaha attorney Greg Lake over a brief citing cases that do not exist. Lake told the justices he accidentally uploaded a working draft from a computer that subsequently malfunctioned and denied using AI. The court referred him for discipline anyway.

In March, a nearly identical scene played out at the Georgia Supreme Court. These are not trial courts in rural counties — these are state supreme courts.

The Pro Se Problem Is Different and Bigger

Beyond the attorney sanctions, the New York Times reported a separate but related crisis: self-represented litigants — people filing without a lawyer — are now using AI to generate their own lawsuits and motions, flooding court dockets with legally incoherent filings.

Courts have historically tried to accommodate pro se filers. That was a reasonable accommodation when the barrier to filing was at least the effort of drafting something yourself. AI eliminates that friction entirely.

Now anyone with a grievance and a chatbot can generate a lawsuit that looks legitimate, complete with case citations that may or may not exist. Clerks still have to process these. Judges still have to read them. Opposing parties still have to respond.

The burden is real, even if the filings are garbage.

Why Is This Still Happening?

"I am surprised that people are still doing this when it's been in the news," Carla Waal, associate dean of information and technology and director of the Gallagher Law Library at the University of Washington School of Law, told NPR.

Waal is developing optional AI ethics training for law students. Optional. At a law school. For something that is currently blowing up courtrooms across the country.

Charlotin's diagnosis is blunter: "We have this issue because AI is just too good — but not perfect."

These tools produce output that sounds authoritative. The citations look real. The legal reasoning sounds coherent. If you do not independently verify every single citation — which takes real work — you will miss the hallucinations. Attorneys under deadline pressure are apparently choosing not to do that work.

What Mainstream Coverage Is Getting Wrong

Both NPR and the New York Times frame this primarily as a story about AI being dangerous or insufficiently regulated. The actual accountability problem runs deeper.

AI did not file these briefs. People did. Lawyers did. They skipped verification. They submitted work product they did not independently check. That is a professional ethics failure.

The rules already exist. Rule 11 of the Federal Rules of Civil Procedure requires attorneys to certify that their filings are accurate to the best of their knowledge. That rule predates AI by decades. The courts are enforcing it. The sanctions are working the way they are supposed to work.

The pro se angle deserves more scrutiny. Both sources mention it, but neither explores the systemic cost — how many judicial hours, clerk hours, and opposing counsel hours are being consumed by AI-generated nonsense filings from people who have no professional accountability.

At least attorneys face disbarment. A random person who uses ChatGPT to sue their landlord over a hallucinated legal theory faces a dismissal. And can refile.

What This Means for Regular People

If you are involved in any litigation right now — as a party, a business, anyone — the legal system you are relying on is slower and more expensive because of this problem.

Court resources are finite. Every hour a judge spends sorting through AI-fabricated nonsense is an hour not spent on real cases with real stakes.

The technology is not going away. The solution is straightforward: verify your citations before you file them, the same way professionals have always been expected to verify their work.

Courts are starting to price that discipline accordingly.