Benjamin v. Costco Wholesale Corp.

Benjamin v. Costco Wholesale Corp., U.S. District Court for the Eastern District of New York, April 24, 2025, 2025 WL 1195925, No. 2:24-cv-07399 (Magistrate Judge Lee G. Dunst). Copy of primary document: full-text reproduction on FindLaw (Primary document not publicly available: https://caselaw.findlaw.com/court/us-dis-crt-ed-new-yor/117206693.html). CourtListener and RECAP were unreachable at verification time.

Victoria Hovsepyan used the AI platform ChatOn, which she had never before used for legal work, to draft a reply in support of a motion to remand. She pasted ChatOn's output into a sworn affirmation submitted under penalty of perjury, spent "maybe twenty minutes" reviewing it, and made no effort to read the cases ChatOn created or to cite check them. Five of the seven cases cited did not exist; the arguments in the reply even undermined her own remand motion.

Violation of Federal Rule of Civil Procedure 11(b)(2); the court found subjective bad faith because filing a sworn affirmation asserting the truth of contents she did nothing to verify was a false statement of knowledge. Sanctions: a $1,000 penalty payable to the EDNY Clerk within 14 days (expressly not attributable to plaintiff Benjamin), and service of the order on her client with proof of service filed on ECF within 14 days. The court declined to refer her to a disciplinary body or revoke her pro hac vice status, citing her candor, remorse, and self-initiated AI-related CLE. Note: the candidate row's metadata mentioned a "public reprimand," but the order imposes no reprimand.

R2, Owned It Late (the court's Order to Show Cause, not the lawyer or opposing counsel, surfaced the problem; her sworn response conceded the fake cases and the ChatOn use; she had completed two AI CLE courses and enrolled in more by the time of the order; the order credits candor and remorse as mitigation for a $1,000 sanction below the cited $1,500–$5,000 range; graded from the full order text, SET-B, full text on a secondary host (FindLaw)).

Never paste AI output into a sworn filing without reading it. A twenty-minute skim of a reply brief is not a reasonable inquiry under Rule 11. If a deadline is tight, seek an extension rather than substituting AI for legal judgment.

• caselaw.findlaw.com: https://caselaw.findlaw.com/court/us-dis-crt-ed-new-yor/117206693.html