ChatGPT, filed by counsel without checking the client's citation.
Andrew Isaacs sued Novartis Pharmaceuticals, claiming he was wrongfully denied severance after his termination. Novartis removed the case to federal court and argued the claims were preempted by ERISA. In a response letter seeking remand, plaintiff's counsel Brian M. Cige cited two cases, Mitchell v. Gilmore and Feliciano v. Riverbay Corp., that do not exist.
The Court could not locate either case and ordered Cige to produce copies. He then confirmed that the cases were not real, took full responsibility, and explained that his client had found them using ChatGPT and that he filed them without checking their accuracy.
Offered for: a procedural-defect remand against Novartis
The Court found that Cige violated Rule 11 by relying on non-existent cases, but it imposed no monetary sanction at that time. Noting that he took immediate responsibility and had already registered for a legal seminar on AI, the Court directed him to certify within 30 days that he completed that October 7, 2025 seminar.
Kingsfield returned REJECT. No case so styled exists. The court could not locate it. Counsel admitted the citation came from the client via ChatGPT.
This matter is among the US case-law cases we ran Kingsfield against in the Charlotin benchmark. Kingsfield's existence check confirms every citation against the case-law corpus and an independent source (CourtListener) before any claim resting on it can stand. Across the run Kingsfield ruled 696 of 700 testable claims correctly, including every one of the 504 fabricated citations and 92 fabricated quotes.
Honest scope. This measures Kingsfield's ruling layer given an extracted citation, not an end-to-end run over a full brief. The case record and the AI-tool attribution come from Damien Charlotin's public database and the court order; the verdict is Kingsfield's. See the full 700-claim benchmark.