A federal judge in Arizona dismissed an employment-discrimination complaint against the U.S. Department of Agriculture after concluding that the 67-page filing displayed familiar signs of artificial-intelligence drafting. Apparently “please generate one lawsuit, extra long” is not yet a recognized rule of civil procedure.
U.S. District Judge Krissa Lanham described the filing as an improper “shotgun pleading”: repetitive, sprawling and unclear about how its facts connected to specific legal claims. The plaintiff, Shelly George, represented herself and said she plans to amend and refile. The dismissal does not decide the ultimate merits of every allegation; it says the current complaint does not present them in a legally usable form.
Lanham barred George from using AI to draft an amended complaint. She also ordered both sides to disclose any future AI assistance and personally verify that every factual and legal assertion is accurate. That distinction matters. The order is not a mystical declaration that software has contaminated the courthouse. It addresses responsibility: a person who files a document remains accountable for what the document says.
Courts have increasingly confronted briefs containing invented cases, distorted quotations and paragraphs that sound authoritative until someone checks them. Some judges have imposed sanctions; others require disclosure. The shared message is less dramatic than “lawyers versus machines.” It is that automation does not move the duty of candor to a server farm.
The case also highlights an access-to-justice problem. Self-represented litigants may turn to generative tools because legal help is expensive and court rules are dense. But a system that produces confident clutter can make a difficult case harder to understand. AI can help organize information; it cannot sign the pleading, answer the judge or take the consequences when the citations lead into the void.
What is verified
This report separates documented facts from commentary and speculation. Source links are listed below for readers who want the original reporting.
Sources
Facts first. Side-eye included.
DJF separates what is confirmed from what is claimed—and tells you why this particular mess is worth your time.