Attorney AI use
In re Sanctions Order of Kenney
The lawyer used one AI tool to check another; the appellate court held a fictitious citation was prima facie evidence of an objectively unreasonable filing.
Sanctions affirmed01
Case background
Material facts
Counsel filed an opposition citing two purported Louisiana appellate decisions that did not exist: Burns v. State, 813 So. 2d 574 (La. App. 1 Cir. 2001), and Smith v. Christus St. Patrick Hosp., 20-306 (La. App. 3 Cir. 11/3/20), 308 So. 3d 1216. (Slip op. at 2.) When defense counsel requested copies, a paralegal supplied AI-generated summaries. After defense counsel asked for PDFs, counsel filed a supplemental memorandum that omitted the two authorities without clearly identifying the errors and introduced additional citation problems. Counsel later said that staff had cross-checked the citations through Google, ChatGPT, and Microsoft Copilot rather than reading the decisions in an authoritative database.
02
The decision
The court’s ruling
The court affirmed $1,368 in fees and costs, required three hours of generative-AI ethics education, ordered a disciplinary report, and assessed appellate costs. Intent or bad faith was not required.
Why it matters
What the decision means
AI cannot be the independent source used to verify AI. As the court stated, “Reviewing AI-generated case summaries is not the same as reading legal cases.” Verification means opening the actual decision, confirming the quotation and proposition, and checking that the authority remains good law.
Primary reading