Filed via 'Federally Lawyer,' an AI legal tool. An invented case plus misplaced quotes.
Jodell Dodge sued FirstService Residential Arizona over employment claims in the District of Arizona. After the defense moved to dismiss her amended complaint, her attorney Matthew Moosbrugger filed a late response brief. The court granted the motion to dismiss, and while reviewing the response it found a string of citation problems.
In all the court counted nine citation defects in the response brief: two misrepresented holdings, five misquoted cases, and one case that does not exist at all, "Gestalt v. City of Gloucester, 2021 WL 4169431 (D. Mass. 2021)." Several real cases were quoted for language that never appears in them, with pincites pointing to the wrong pages.
At a show-cause hearing, Moosbrugger testified under oath that he used an AI tool called Federally Lawyer, an add-on to the enterprise version of ChatGPT. Because he filed the brief late, he admitted he did not independently review any citations and only asked ChatGPT whether the cases were real and still good law.
Offered for: supporting authority in the brief
The court found Moosbrugger submitted non-existent caselaw and misrepresented holdings, but decided severe sanctions were not warranted. It credited his honest admission, his self-report to the State Bar of Arizona, his contingency-fee representation with no fee collected, and the public humiliation already involved. The court imposed no further sanctions, encouraged him to attend AI-focused legal education, and directed the clerk to send copies of the order to the State Bar of Arizona Disciplinary Committee.
Kingsfield returned REJECT. The court found the cited case does not exist. Other quotations in the same brief were tied to pages that do not support them.
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.