Skip to content

Courts on AI / Fabricated or misquoted citations

Ponder v. BCG Equities, LLC

United States District Court for the Eastern District of Michigan · Federal district courts · MI · · No. 2:25-13474 · order

The Eastern District of Michigan resets a pro se FCRA case by striking pending motions and setting aside default. The order cautions both sides after a brief cites Antoine for a quotation and proposition not found there, warning that generative-AI phantom cases can trigger Rule 11 sanctions.

Disposition
The court struck pending motions, set aside default, set a responsive-pleading deadline, and warned both parties about generative AI.
Recorded conduct
A party cited Antoine for a quotation and proposition that the court says do not appear at the cited source, suggesting possible generative-AI use. — outcome: warning; actor: other
Canonical record
Incident tracker entry — verified outcome, regulator disposition and related rules

What the court wrote

Verbatim from the court's document (public domain). Ellipses mark omitted text; nothing is paraphrased.

Finally, the Court issues a caution about the use of generative artificial intelligence (“AI”). It appears from the briefing that one or both parties may be utilizing AI to draft their briefs. (See, e.g., ECF No. 20, PageID.71 (citing Antoine v. Atlas Turner, Inc., 66 F.3d 105, 110 (6th Cir. 1995) for a quotation and proposition that does not appear at the citation).) The parties should take note: AI cannot give legal advice, and, in many cases, its use in legal research has led to glaring misstatements of law and related sanctions awarded against both attorneys and pro se litigants. “At this point, it ‘is no secret that generative AI programs are known to ‘hallucinate’ nonexistent cases, and with the advent of AI, courts have seen a rash of cases in which both counsel and pro se litigants have cited such fake, hallucinated cases in their briefs.’” Seither & Cherry Quad Cities, Inc. v. Oakland Automation, LLC, No. 23-11310, 2025 WL 2105286, at *1 (E.D. Mich. July 28, 2025) (Behm, J.) (quoting Sanders v. United States, 176 Fed. Cl. 163, 169 (2025)). AI chatbots “are designed to mimic patterns of words, probabilistically. When they are ‘right,’ it is because correct things are often written down in the dataset they were trained on, not because they can independently fact-check themselves in the same way a human would.” Id. Indeed, French data scientist Damien Charlotin catalogued “at least 490 court filings” between May and October 2025 that contained hallucinations. ... Thus, the use of generative AI when drafting legal filings is a very risky practice, as pro se litigants and attorneys are still subject to the requirements of Fed. R. Civ. P. 11(b)(2) with respect to pleadings, motions and other documents they sign and submit to the Court; even in the absence of bad faith, Rule 11 sanctions for use of AI generated phantom cases may be warranted. ... The parties are HEREBY SO WARNED.

Authorities cited in the passage: Antoine v. Atlas Turner, Inc., 66 F.3d 105 (6th Cir. 1995) · Seither & Cherry Quad Cities, Inc. v. Oakland Automation, LLC, 2025 WL 2105286 (E.D. Mich. July 28, 2025) · Sanders v. United States, 176 Fed. Cl. 163 (2025) · Ali v. IT People Corp., Inc., 2025 WL 2682622 (E.D. Mich. Sept. 19, 2025) · Fed. R. Civ. P. 11(b)(2)

compiled by the Cognesio LLP editorial teammethodas of 2026-09-07v0.1.1coding CC BY 4.0JSON Hugging Face GitHub