915 Opinions Later, "I Didn't Know" No Longer Works

On April 28, 2026, a federal judge in the Central District of California sanctioned an attorney for submitting a fabricated case citation three times across two separate briefs, and for his repeated attempts to minimize and conceal the conduct afterward. (Chaney v. Transdev Services Inc. et al., No. 2:24-cv-10761-ODW, C.D. Cal., Apr. 28, 2026, at 10–12.)

The tool involved was not ChatGPT. It was LexisNexis+ and its drafting feature, Protégé, a purpose-built legal research product. (Id. at 9.) If you have been operating under the assumption that commercially marketed legal AI is safer than general-purpose tools, this order is worth reading carefully.

The court's bad faith finding rests on two grounds. First, the court cited 915 legal opinions addressing improper AI use in the past three years and held that any attorney fulfilling the duty to stay current on technology under Rule 1.1 is already on notice of the risk. Claiming ignorance is no longer available. (Id. at 10.) Second, the concealment pattern was treated as an independent basis for sanctions. The attorney framed the citation as a copy-and-paste error, admitted AI use only after a show cause order, and never voluntarily disclosed that the same fabricated case appeared twice more in a second brief filed the same day. (Id. at 10–11.)

The financial sanction was $2,500. The more significant one: counsel must file a declaration in every active Central District case explaining his AI use and the court's bad faith findings, and must email that declaration to the chambers of every C.D. Cal. judge before whom he is not currently appearing. (Id. at 11–12.)

Orthodoxy was built for the verification step courts are now treating as a professional obligation. The record it creates is the record that did not exist here.